OXFORD

APPLIED
LINGUISTICS

The Study of
Second Language
Acquisition
Rod Ellis

5. vv
■ypEAKV

WP 2210045 8
The Study of Second
Language Acquisition
Rod Ellis
3

Oxford University Press

Oxford University Press
Great Clarendon Street, Oxford, 0X2 6DP

Oxford New York

Athens Auckland Bangkok Bogota Bombay Buenos Aires
Calcutta Cape Town Dar es Salaam Delhi Florence Hong Kong
Istanbul Karachi Kuala Lumpur Madras Madrid Melbourne
Mexico City Nairobi Paris Singapore Taipei Tokyo Toronto Warsaw

and associated companies in

Berlin Ibadan

Oxford and Oxford English are trade marks of

Oxford University Press.

ISBN 0 19 437189 1

© Rod Ellis 1994

First published 1994

Fifth impression 1997

No unauthorized photocopying

All rights reserved. No part of this publication may be reproduced,

stored in a retrieval system, or transmitted, in any form or by any
means, electronic, mechanical, photocopying, recording or otherwise,
without the prior written permission of Oxford University Press.

This book is sold subject to the condition that it shall not, by way of
trade or otherwise, be lent, resold, hired out, or otherwise circulated
without the publisher’s prior consent in any form of binding or cover
other than that in which it is published and without a similar
condition including this condition being imposed on the subsequent
purchaser.

Set by Tradespools Ltd., Frome, Somerset

Printed in Hong Kong

Contents

Acknowledgements v

Introduction \

PART ONE
Background 7

Introduction 9

1 Second language acquisition research: an overview 11

PART TWO
The description of learner language 41

Introduction 43

2 Learner errors and error analysis 47

3 Developmental patterns: order and sequence in

second language acquisition 73

4 Variability in learner language 119

5 Pragmatic aspects of learner language 159

PART THREE
Explaining second language acquisition: external factors 191

Introduction 193

6 Social factors and second language acquisition 197

7 Input and interaction and second language acquisition 243

PART FOUR

Explaining second language acquisition: internal factors 293

Introduction 295

8 Language transfer 299

 9 Cognitive accounts of second language acquisition 347

10 Linguistic universal and second language acquisition 415

PART FIVE

Explaining individual differences in second language

acquisition 467

Introduction 469

11 Individual learner differences 471

12 Learning strategies 529

PART SIX

Classroom second language acquisition 561

Introduction 563

13 Classroom interaction and second language acquisition 565

14 Formal instruction and second language acquisition 611

PART SEVEN

Conclusion 665

Introduction 667

15 Data, theory, and applications in second language

acquisition research 669

Glossary 692

Bibliography 730

Author index 792

Subject index 805

Acknowledgements

Many people have assisted me in the preparation of this book, but, of course,
I alone am responsible for any shortcomings. I would like to give special
thanks to Henry Widdowson and Gaby Kasper, both for their support and
their critical acumen. Others who have commented on individual chapters
are Leslie Beebe, Kevin Gregg, Eric Kellerman, Patsy Lightbown, Barry
McLaughlin, Teresa Pica, Peter Skehan, Elaine Tarone, Leo van Lier, Lydia
White, and Julia Sallabank of Oxford University Press. Above all I am grate¬
ful for the support provided by Temple University Japan: both its students,
who have been instrumental in shaping the contents of the book, and Ken
Schaefer for helping me to maintain belief in its value and for ensuring that I
had the time to write it. A special word of thanks is due to Andrew Jones for
helping me to prepare the manuscript for publication and for providing a
final—and crucial—‘test’ of its contents.
The author and publisher are grateful to the following for permission to re¬
produce extracts and figures from copyright material:
Ablex Publishing Corporation: extract from Blum-Kulka, S., House, J., and
Kasper, G.: Cross Cultural Pragmatics (1989), and table from Pica, T. and
Doughty C., ‘Variations in classroom interaction as a function of participa¬
tion pattern and task’ in Fine, J. (ed.): Second Language Discourse: A
Textbook of Current Research (1988);
Academic Press Inc: table from Givon, T.: On Understanding Grammar
(1979), and a figure from Wenk, B. and Wioland, F., ‘Is French really
syllable-timed?’ in Journal of Phonetics: 10 (1982);
John Benjamins Publishing Company, Amsterdam/Philadelphia: figures from
Lydia White: Universal Grammar and Second Language Acquisition
(1989), and table from Susan Gass and Josh Ard, ‘Second Language
acquisition and the ontology of language universal’ in William E.
Rutherford (ed.): Typological Universals and Second Language Acquisi¬
tion (1984);
Blackwell Publishers and the author: figure from Preston, D.R.: Sociolinguis¬
tics and Second Language Acquisition (1989);
Cambridge University Press and the authors: table from Bickerton, D.: Dy¬
namics of a Creole System (1975), table from Stern, H., ‘Analysis and ex¬
perience as variables in second language pedagogy’ in Harley, B. et ah: The
Development of Second Language Proficiency (1990), figure from All-
wright, D. and Bailey, K.: Focus on the Language Classroom: An Introduc-
vi Acknowledgements

tion to Classroom Research for Language Teachers (1991), table from

Young, R., ‘Variation and the interlanguage hypothesis’ in Studies in Se¬
cond Language Acquisition 10 (1988), table from Kellerman, E., ‘Transfer
and non-transfer: Where are we now?’ in Studies in Second Language Ac¬
quisition 2 (1979), table from Gardner, R. and McIntyre, P., ‘An instru¬
mental motivation in language study: Who says it isn’t effective?’ in Studies
in Second Language Acquisition 13 (1991), and adapted figure from
White, L., ‘Argument structure in second language acquisition’ in Journal
of French Language Studies (1991);
Edinburgh University Press: table from Long, M. and Sato, C., ‘Methodolo¬
gical issues in interlanguage studies: an interactionist perspective’ in
Davies, A. et ah: Interlanguage (1984);
Professor Robert C. Gardner: figure from Gardner, R.: Social Psychology and
Second Language Learning: The Role of Attitude and Motivation
(Edward-Arnold, 1985);
Heinle and Heinle Publishers: table from Gass, S. and Varonis, E., ‘Task vari¬
ation and nonnative/nonnative negotiation of meaning’, and extract
from Olshtain, E. and Blum-Kulka, S., ‘Degree of approximation: non¬
native reactions to native speech act behavior’, both in Gass, S. and
Madden, C. (eds.): Input in Second Language Acquisition (Newbury
House, 1985), and figure from Oxford, R.: Language Learning Strategies:
What Every Teacher Should Know (Newbury House, 1990);
ICG Publication: figure from Chomsky, N.: Lectures on Government and
Binding (Foris, 1981);
A. Jones: table from unpublished paper, ‘Review of studies on the acquisition
of relative clauses in English’ (1991);
Professor G. Kasper: diagram from Kasper, G., ‘Perspectives on language
transfer’ in BAAL Newsletter 24 (1984);
The Macmillan Press Ltd: table from Nunan, D., ‘The teacher as researcher’
in Brumfit, C. and Mitchell, R. (eds.): Research in the Language Class¬
room, ELT Documents No. 133 (Modern English Publications, in associ¬
ation with the British Council, 1990);
Oxford University Press: figure from Pit Corder, S.: Error Analysis and Inter¬
language (1981), 2 tables from Ellis, R.: Understanding Second Language
Acquisition (1985), figure from Tarone, E., ‘On the variability of inter¬
language systems’ in Applied Linguistics 4 (1983), figure from Bialystok, E.
and Sharwood Smith, M., ‘Interlanguage is not a state of mind: an
evaluation of the construct for second language acquisition’ in Applied
Linguistics 6 (1985), and table from Mangubhai, F., ‘The processing
behaviours of adult second language learners and their relationship to
second language proficiency’ in Applied Linguistics 12 (1991);
Prentice-Hall Inc., Englewood Cliffs, New Jersey: table from Chamot, A.,
‘The learning strategies of ESL students’ in Wenden, A. and Rubin, J. (eds.):
Learner Strategies in Language Learning © 1987;
 Acknowledgements vii

The Research Club in Language Learning and the authors: table from Schach-
ter, J., ‘An error in error analysis’ (1974), in Language Learning 27, extract
from Eckman, F., ‘Markedness and the contrastive analysis hypothesis’
(1977), in Language Learning 27, figure from Bialystok, E., ‘A theoretical
model of second language learning’ (1978), in Language Learning 29, fig¬
ure from Zobl, H., ‘Markedness and the projection problem’ (1983) in
Language Learning 33;
Sage Publications Inc.: figure from Gass, S. and Varonis, E., ‘Miscommun-
ication in nonnative speaker discourse’ in Coupland, N., Giles, H., and
Wiemann, J. (eds.): Miscommunication and Problematic Talk (1991);
Teachers of English to Speakers of Other Languages Inc. (TESOL) and the au¬
thors: figure from Krashen, S., ‘Some issues relating to the monitor model’
(1977) and table from Tarone, E., ‘Typology of communication strategies
in interlanguage: a progress report’ (1977), both in Brown, H., Yorio, C.
and Crymes, R. (eds.): On TESOL ’77, table from Allen, J.P.B., Frohlich,
M. and Spada, N., ‘The communicative orientation of language teaching:
an observation scheme’ (1984) in Handscome, J., Orem, R., and Taylor, B.
(eds.): On TESOL ’83: The Question of Control, and table from Long, M.,
‘Does second language instruction make a difference? A review of the re¬
search’ (1983) in TESOL Quarterly 17.

Despite every effort to trace and contact copyright holders before publica¬
tion, this has not always been possible. If notified, the publisher will be
pleased to rectify any errors or omissions at the earliest opportunity.
For Takayo
Introduction

Second language acquisition research: some developments

Although it is not possible to set a precise date on when second language ac¬
quisition (SLA) research first established itself as a field of enquiry, there is
general agreement that it took place around the end of the 1960s. At this time,
some of the first studies of second language (L2) learners were published (for
example, Ravem 1968; Huang 1970). At the same time a theoretical case for
examining L2 acquisition was advanced (for example, Corder 1967). From
that point SLA research developed rapidly and continues to do so. There has
been an enormous amount of empirical research directed at describing the
characteristics of L2 learner language and how these change as acquisition
takes place. There has also been a growing interest in theory construction, as
reflected in the plethora of frameworks, models and theories now available.
The developments in SLA research over the years have been of several
kinds. One development concerns the scope of the field of enquiry. Whereas
much of the earlier work focused on the linguistic—and, in particular, the
grammatical—properties of learner language and was psycholinguistic in
orientation, later work has also attended to the pragmatic aspects of learner
language and, increasingly, has adopted a sociolinguistic perspective. Thus,
whereas many researchers continue to focus their attention on how L2 learn¬
ers develop grammatical competence, many others are concerned with how
learners develop the ability to perform speech acts such as requests or apol¬
ogies appropriately. Similarly, whereas many researchers continue the long¬
standing attempt to explain the psycholinguistic processes that underlie L2
acquisition and use, others have given attention to the social factors that in¬
fluence development. The result is that the scope of SLA research is now much
wider than in the 1970s or even early 1980s.
A second development concerns the increasing attention paid by SLA
research to linguistic theory. This is most evident in the studies based on
Chomsky’s model of grammar (Universal Grammar), but is also apparent in
work based on functional models of language (for example, Givon 1979a;
1995) and language typology (for example, Comrie 1984). Thus, whereas
much of the earlier work made use of fairly simple grammatical concepts,
derived from descriptive grammars, the later work makes greater use of
technical concepts derived from a particular theory of grammar. The
relationship between SLA research and linguistics is increasingly symbiotic,
2 Introduction

in the sense that L2 researchers and linguists both draw on each others’ work.
SLA research, therefore, is no longer a consumer of linguistics, but also a
contributor to it.
Another significant development in SLA research has been the increase in
theory-led research, which, in part, is a reflection of its relationship with lin¬
guistics. Much of the earlier research was of the ‘research-then-theory’ kind.
Typically, it involved the collection of samples of learner language, their ana¬
lysis, and the description of their main characteristics. Explanations of the
main features were a matter of post hoc interpretation. This was both inevit¬
able and desirable given the lack of hard information about what learners do
when they acquire an L2. However, as theories have been developed, re¬
searchers have increasingly sought to confirm them by identifying and testing
specific hypotheses, often experimentally. A brief examination of some of the
current journals that publish L2 acquisition research (for example, Language
Learning, Studies in Second Language Acquisition, Second Language Re¬
search) will reveal the current primacy of theory-led, experimental-type re¬
search. However, although there has been a general increase in confirmatory
research, many studies of the interpretative kind continue to be published.
The main thrust of SLA research in the early years addressed the universal
properties of L2 acquisition—looking for commonalities across learners of
different ages, in different settings and with different Lis. There was, how¬
ever, an alternative tradition that pre-dated the early studies of learner lan¬
guage—the study of how individuals differ. The 1950s and early 1960s, for
example, had witnessed intensive research on language aptitude, one of the
main factors contributing to individual differences, while the late 1960s and
1970s also saw a substantial body of work on motivation. This tradition,
which for a while was partially submerged as SLA research established itself,
has now reasserted itself and is viewed as part of the total field of enquiry.
However, the study of L2 acquisition as a phenomenon with universal pro¬
perties and the study of individual differences in L2 learners have not been
fully integrated; the study of ‘learning’ and of ‘learners’, somewhat surpris¬
ingly, exist as separate areas within SLA research.
It is also possible to detect the emergence of another fairly distinct sub-field
of SLA research devoted to investigating classroom L2 acquisition. Class¬
room studies of L2 learners date back to the inception of SLA research.
However, many of the early studies deliberately focused on learners in
natural settings. Subsequently, studies of classroom learners, either in actual
classrooms or in settings that simulate classroom conditions, have prolifer¬
ated. Some of these studies were designed to investigate issues important to
language pedagogy but many were explicitly concerned with theoretical
issues related to L2 acquisition.
As a result of these developments, SLA research has become a rather
amorphous field of study with elastic boundaries. This makes the task of sur¬
veying the field a difficult one, and only a few few attempts have been made to
 Introduction 3

do so: for example, Ellis 1985a, McLaughlin 1987, Spolsky 1989, and
Larsen-Freeman and Long 1991. Instead, scholars have usually preferred to
identify specific areas of SLA research. Thus, books on such topics as learner
errors, vocabulary acquisition, Universal Grammar and L2 acquisition, L2
pragmatics, variability, the role of input, the role of social factors, attitudes
and motivation, learner strategies, and classroom L2 acquisition are more
common. In a sense, SLA research stands at the cross-roads. It may continue
as a coherent field of enquiry with its own recognized research community, or
it may splinter into a series of sub-fields, and, in some cases, perhaps become
submerged in the ‘mother’ disciplines that inform these sub-fields.

The aims of the book

The extent to which SLA research can claim to constitute a discipline in its
own right depends on whether it is possible to identify a defined field of en¬
quiry and a body of knowledge relating to it. The main aim of this book,
therefore, is to develop a framework for describing the field as it currently
exists and to use this framework to provide an extensive account of what is
currently known about L2 acquisition and L2 learners. It is intended to tackle
this ‘descriptively’—to avoid taking up any particular position regarding
what constitutes the most legitimate approach to SLA research. The account
offered, therefore, aims to be balanced and objective, as far as this is possible.
Inevitably, though, the personal views of the author will colour the picture
provided. Also, although the emphasis will be on ‘describing’ what has been
discovered, attention will also be given to the evaluation of the methodologies
that have been used to conduct research, of the significance of the information
that has been unearthed, and of the theories that have been developed to ex¬
plain the findings. Again, though, as far as possible, these evaluations will
strive for objectivity by reflecting a range of opinion.

The book’s audience

This book has been written for three kinds of reader. One is students of SLA
research—those who are beginning their study of L2 acquisition and wish to
obtain an understanding of the principal issues that have been addressed, the
methods used to research them, the main findings and the theories that have
been developed to explain them. It is hoped that the book will provide an ac¬
cessible introduction to the field.
The second kind of reader is the SLA researcher who feels the need for a ref¬
erence book that provides an overview of the main work accomplished in the
different areas of SLA research. Such readers are likely to be experts in one or
more areas of SLA research and, therefore, may have little need for the chap¬
ters that deal with their areas of specialism, but may wish to benefit from a
survey of the work undertaken in other areas with which they have less famili-
4 Introduction

arity. As has already been pointed out, the scope of the field is now very large
and the amount of published research enormous, with the result that few re¬
searchers are able to keep themselves up to date in work undertaken in all
areas.
The third, and probably the principal kind of reader for which this book is
intended, is second/foreign language teachers, many of whom may be com¬
pleting a post-service programme of teacher education (for example, an MA
or Diploma in TESOL or Applied Linguistics). It is pertinent, therefore, to
consider why teachers should make the effort to read about L2 acquisition.

Second language acquisition research and language pedagogy

The main reason why language teachers need to be familiar with SLA re¬
search was stated in my earlier book (Ellis 1985a) as follows:

... unless we know for certain that the teacher’s scheme of things really
does match the learner’s way of going about things, we cannot be sure that
the teaching content will contribute directly to language learning (Ellis
1985a: 1).

The study of SLA provides a body of knowledge which teachers can use to
evaluate their own pedagogic practices. It affords a learning- and learner-
centred view of language pedagogy, enabling teachers to examine critically
the principles upon which the selection and organization of teaching have
been based and also the metholodogical procedures they have chosen to em¬
ploy. Every time teachers make a pedagogic decision about content or meth¬
odology, they are, in fact, making assumptions about how learners learn. The
study of SLA may help teachers in two ways. First, it will enable them to make
their assumptions about learning explicit, so that they can examine them crit¬
ically. In this way, it will help them to develop their own explicit ideas of how
the kind of learners they are teaching acquire an L2. Second, it will provide
them, information that they can use when they make future pedagogic
decisions.
Of course, SLA research is not the only source of information of relevance
to language teachers. There are other areas of enquiry—some with much lon¬
ger histories than SLA research—which are important. Stern (1983) identifies
five areas that language pedagogy draws on for its ‘fundamental con¬
cepts’—the history of language teaching, linguistics, sociology, psychology,
and education. SLA research, as we have seen, is, like language pedagogy it¬
self, a hybrid discipline, drawing on a range of other disciplines for its own
constructs. However, it would be a mistake to treat SLA research as a me¬
diating discipline that takes concepts from other disciplines and moulds them
into a form applicable to language pedagogy. SLA research has its own
agenda and is best treated as another source discipline. Language pedagogy
will draw on SLA research in the same way as it draws on other areas of en-
 Introduction 5

quiry. If it is given privileged status, then, this will only be because its
aims—the description and explanation of how L2 learning takes place—are
of obvious relevance to the practice of language teaching.
The information provided by SLA research, then, needs to be ‘applied’ in
the same way as that from other sources. Indeed, caution needs to be exer¬
cised when it comes to making use of the information it supplies. SLA re¬
search is not capable of providing teachers with recipes for successful
practice. As will become clear later, there is no comprehensive theory of SLA,
nor even any single theory that is widely accepted. Researchers do not even
agree completely over what constitutes the descriptive ‘facts’ of L2 acquisi¬
tion. There is much disagreement and controversy—a reflection of the highly
complex nature of L2 acquisition and the relative immaturity of the field. For
these reasons, SLA research should be treated as providing teachers with ‘in¬
sights’ which they can use to build their own explicit theory. It is on the basis
of this theory—not on the basis of SLA research itself or any theory it has pro¬
posed—that teaching practice should proceed.

The structure and contents of the book

The contents of this book have been organized in accordance with a general
conceptual framework that takes account of (1) a general distinction between
the ‘description’ and ‘explanation’ of L2 acquisition, and (2) the various sub¬
fields that have developed over the years. The following are the main Parts
with a brief summary of their contents:

Part 1: Background
This section contains one chapter that outlines the conceptual framework of
the whole book.

Part 2: The description of learner language

This section reports some of the main findings regarding the nature of learner
language. It considers learner errors, developmental patterns, variability, and
pragmatic features.

Part 3: Explaining second language acquisition:

external factors
This section begins the task of explaining L2 acquisition by considering
external influences—the role of social factors and of input/interaction.

Part 4: Explaining second language acquisition: internal factors

This section continues the work of explaining L2 acquisition by examining
6 Introduction

various theories of the mental processing that learners engage in.

Three perspectives on these processes are offered—language transfer, cog¬
nitive accounts of L2 acquisition, and linguistic universals.

Part 5: Explaining individual differences in second language

acquisition
In this section the focus of attention switches from ‘learning’ to ‘the learner’.
Individual differences are considered from the point of view of general
psychological factors (for example, motivation) and learner strategies.

Part 6: Classroom second language acquisition

This Part examines classroom-based and classroom-orientated research,
from the point of view of both interaction and formal instruction.

Part 7: Conclusion
The book concludes with a critical look at the ‘state of the art’ of SLA research
from the point of view of the data it works with, theory construction, and its
applications.
There is no separate chapter on research methodology in SLA research,
mainly because the methods used vary considerably according to the particu¬
lar aspect of SLA being studied. However, where appropriate, information
about the methods used to investigate specific areas is provided in the indi¬
vidual chapters.

A note on terminology
The term ‘second language acquisition research’ (SLA research) is used to re¬
fer to the general field of enquiry. It labels the discipline that is the focus of this
book. The term ‘L2 acquisition’ serves as an abbreviation for ‘the acquisition
of a second language’. This is what learners try to do and is the object of study
in SLA research.
For reasons explained later no distinction is made between ‘acquisition’
and ‘learning’, the two terms being used interchangeably.
Throughout the book, words explained in the Glossary are in italic type.
PART ONE

Background
Introduction

As was noted in the Introduction to this book, second language acquisition

research (SLA research) has developed into a wide-reaching and somewhat
amorphous field of enquiry, drawing on and contributing to a number of dis¬
tinct disciplines—linguistics, cognitive psychology, psycholinguistics, socio¬
linguistics, and education. The extent to which it is still justifiable to treat SLA
research as a coherent field, with identifiable goals and methods of enquiry,
depends on whether it is possible to provide a clear definition of the object of
enquiry (second language acquisition) and to delineate a framework that
gives some structure to the issues it seeks to address.
The purpose of the chapter in this section (Chapter 1) is to attempt such a
definition and to outline a framework that can serve as a means for examining
the key issues. It asserts that the study of second language acquisition entails
both the description of learner language, as it develops over time, and the ex¬
planation of its characteristics. While a strict separation of these twin
goals—description and explanation—is not possible, it provides a way into
the study of a highly complex field. It also accords with the way the field has
developed historically, for whereas much of the earlier work concentrated on
description, the later wiork has been more concerned, perhaps, with the
search for a theory of second language acquisition.
 ' hrP* '

'

' ,
1 Second language acquisition research:
An overview

Introduction

The main purpose of this chapter is to set the scene for the rest of the book. In
order to do this the chapter will (1) examine what is meant by the term Second
Language Acquisition, (2) identify a number of central questions which
second language acquisition research has addressed, and (3) present a frame¬
work for examining four major areas of enquiry and provide a brief survey of
work done in each.

What is second language acquisition?

The obvious first step in the exploration of SLA research is to establish a clear
understanding of what the object of the field of study is—second language ac¬
quisition. This is particularly important because, as will become evident in
subsequent chapters, the nature of this object is far from clear, and different
researchers have given very different interpretations of it. We will begin by
examining a number of points relating to the meaning of the term ‘second lan¬
guage acquisition’.

‘Second’ v. ‘third’ language acquisition

Many learners are multilingual in the sense that in addition to their first lan¬
guage they have acquired some competence in more than one non-primary
language. Multilingualism is the norm in many African and Asian countries.
Sometimes a distinction is made between a ‘second’ and a ‘third’ or even
‘fourth’ language. However, the term ‘second’ is generally used to refer to any
language other than the first language. In one respect this is unfortunate, as
the term ‘second’ when applied to some learning settings, such as those in
South Africa involving black learners of English, may be perceived as op¬
probrious. In such settings, the term ‘additional language’ may be both more
appropriate and more acceptable.

Second v. foreign language acquisition

A distinction between second and foreign language acquisition is sometimes

made. In the case of second language acquisition, the language plays an
12 Background

institutional and social role in the community (i.e. it functions as a recognized

means of communication among members who speak some other language as
their mother tongue). For example, English as a second language is learnt in
the United States, the United Kingdom, and countries in Africa such as
Nigeria and Zambia. In contrast, foreign language learning takes place in
settings where the language plays no major role in the community and is
primarily learnt only in the classroom. Examples of foreign language learning
are English learnt in France or Japan.
The distinction between second and foreign language learning settings may
be significant in that it is possible that there will be radical differences in both
what is learnt and how it is learnt. However, for the time being the extent to
which the sociolinguistic conditions of learning determine learning outcomes
or learning processes must remain an open question—to be answered as a res¬
ult of our exploration of SLA research. This is addressed in greater depth in
Chapter 6. There is a need for a neutral and superordinate term to cover both
types of learning. Somewhat confusingly, but in line with common usage, the
term ‘second language acquisition’ will be used for this purpose.

Naturalistic v. instructed second language acquisition

A distinction will be made between naturalistic and instructed second lan¬

guage acquisition, according to whether the language is learnt through com¬
munication that takes place in naturally occurring social situations or
through study, with the help of ‘guidance’ from reference books or classroom
instruction.
Klein (1986) similarly distinguishes ‘spontaneous’ and ‘guided’ acquisi¬
tion, treating the distinction as a psycholinguistic one. He argues that the
learner focuses on communication in naturalistic second language acquisi¬
tion and thus learns incidentally, whereas in instructed second language
acquisition the learner typically focuses on some aspect of the language
system. It may be better, however, to view the distinction as only a socio¬
linguistic one—i.e. reflecting the settings and activities in which learners
typically participate. It would certainly be wrong to assume that naturalistic
learning is subconscious and instructed learning conscious. It remains an
open question as to whether the process of acquisition is the same or different
in naturalistic or classroom settings. In this book, therefore, the distinction
will be used only in its sociolinguistic sense.

Competence v. performance

A distinction is often made between linguistic competence and performance

when studying language. According to Chomsky (1965), competence con¬
sists of the mental representations of linguistic rules that constitute the
speaker-hearer’s internal grammar. This grammar is implicit rather than
 Second language acquisition research: An overview 13

explicit and is evident in the intuitions which the speaker-hearer has about the
grammaticality of sentences. Performance consists of the use of this grammar
in the comprehension and production of language. The distinction between
competence and performance has been extended to cover communicative as¬
pects of language (see Hymes 1971a; Canale and Swain 1980). Communicat¬
ive competence includes knowledge the speaker-hearer has of what
constitutes appropriate as well as correct language behaviour and also of
what constitutes effective language behaviour in relation to particular com¬
municative goals. That is, it includes both linguistic and pragmatic know¬
ledge. Communicative performance consists of the actual use of these two
types of knowledge in understanding and producing discourse.
The main goal of SLA research is to characterize learners’ underlying
knowledge of the L2, i.e. to describe and explain their competence. However,
learners’ mental knowledge is not open to direct inspection; it can only be in¬
ferred by examining samples of their performance. SLA researchers have used
different kinds of performance to try to investigate competence. Many ana¬
lyse the actual utterances that learners produce in speech or writing (for ex¬
ample, Larsen-Freeman 1975). Some try to tap learners’ intuitions about
what is correct or appropriate by means of judgement tasks (for example,
White 1985), while others rely on the introspective and retrospective reports
that learners provide about their own learning (for example, Cohen 1984).
Needless to say none of these provide a direct window into competence. Also,
not surprisingly, very different results can be obtained depending on the kind
of performance data the researcher studies.

Usage v. use
The distinction between usage and use was first proposed by Widdowson
(1978) to facilitate a discussion of language pedagogy but it is equally applic¬
able to language acquisition. Usage is ‘that aspect of performance which
makes evident the extent to which the language user demonstrates his know¬
ledge of linguistic rules’ (p. 3). We study usage if we focus attention on the
extent to which the learner has mastered the formal properties of the
phonological, lexical, and grammatical systems. Use is that aspect of per¬
formance which ‘makes evident the extent to which the language user demon¬
strates his ability to use his knowledge of linguistic rules for effective
communication’. We study use if we examine how learners convey meaning
through the process of constructing discourse. One way in which this can be
undertaken is by studying pragmatic aspects of language, such as how
learners learn to perform speech acts like requests and apologizing.
SLA research has been primarily concerned with studying usage, although
it is now paying more attention to use. It has become apparent that even if the
aim is to find out how learners acquire purely formal features (such as verb +
-ing or copula ‘be’), it is often necessary to examine how they use these
14 Background

features to express meaning. Thus, analyses based on ‘form’ only have

increasingly given way to analyses of ‘form-function’ and ‘function-form’
correspondences. Many researchers are no longer content to ask whether
learners provide evidence of the correct usage of a form such as the present
progressive tense in contexts that call for it. They also want to know what
meanings learners use the form to perform and, in particular, whether they
use it to perform meanings other than those it serves in the target language.
For example, learners who say ‘Sharpening please’ when they want someone
to sharpen their pencil are using verb + -ing to perform a function (command¬
ing) which it does not usually have in the target language. Other researchers,
particularly those interested in pragmatic characteristics of learner language,
take ‘function’ as their starting point and investigate the different forms that
learners use to perform specific functions.
Precisely what should be the object of enquiry in SLA research is a matter of
controversy, however. On the one hand, there are researchers who argue that
many properties of language are purely formal in nature.1 On the other, there
are those who emphasize the importance of studying the functional uses of
language. SLA research currently employs both types of enquiry.

‘Acquisition’

In order to study how learners acquire a second language, a clear, operational

definition of what is meant by the term ‘acquisition’ is needed. Unfortunately,
researchers have been unable to agree on such a definition. ‘Acquisition’ can
mean several things.
First, some researchers (for example, Krashen 1981) distinguish between
‘acquisition’ and ‘learning’. The former refers to the subconscious process of
‘picking up’ a language through exposure and the latter to the conscious pro¬
cess of studying it. According to this view, it is possible for learners to ‘ac¬
quire’ or to ‘learn’ rules independently and at separate times. Although such a
distinction can have strong face validity—particularly for teachers—it is
problematic, not least because of the difficulty of demonstrating whether the
knowledge learners possess is of the ‘acquired’ or ‘learnt’ kind.2 In this book
the terms ‘acquisition’ and ‘learning’ will be used interchangeably. They will
be placed inside inverted commas if used in their distinctive senses.
Second, researchers disagree about what kind of performance they think
provides the best evidence of acquisition. We have already noted that some
researchers work with production data, some study learners’ intuitions about
the L2, while others access learners’ introspections. Also, some researchers
(for example, Bickerton 1981) consider a feature has been acquired when it
appears for the first time, while others (for example, Dulay and Burt 1980) re¬
quire the learner to use it to some predetermined criterion level of accuracy,
usually 90 per cent. Thus, a distinction can be made between acquisition as
‘emergence’ or ‘onset’ and acquisition as ‘accurate use’.
 Second language acquisition research: An overview 15

Clearly ‘acquisition’ can mean several very different things. This makes it
very difficult to compare the results of one study with those of another. Con¬
flicting results can be obtained depending on whether the data used consist of
learners’ productions, introspections, or intuitions, or whether emergence or
accuracy serves as the criterion of acquisition. It is for this reason that it is im¬
portant to examine carefully the nature of the data used and the way in which
acquisition has been measured, when reading reports of actual studies.

Summary
There is no simple answer to the question ‘What is second language acquisi¬
tion?’ It can take place in either a naturalistic or an instructional setting, but
may not necessarily differ according to the setting. The goal of SLA is the
description and explanation of the learner’s linguistic or communicative
competence. To this end, the researcher must examine aspects of the learner’s
usage or use of the L2 in actual performance, by collecting and analysing
either samples of learner language, reports of learners’ introspections, or
records of their intuitions regarding what is correct or appropriate L2 behavi¬
our. The acquisition of an L2 feature may be considered to have taken place
either when it is used for the first time or only when it can be used to a high
level of accuracy.
Second language acquisition is a complex, multifaceted phenomenon and it
is not surprising that it has come to mean different things to different people.

General questions in second language acquisition research

The study of how learners learn a second language does not have a very long
history. The surge of empirical work that informs current thinking did not
begin until the late 1960s. Although there have been several de¬
velopments in SLA since its inception, it is still possible to identify a set of
general questions for which researchers have sought answers.

1 What do second language learners acquire?

The first question, which guided much of the early research in the late sixties
and seventies, concerned what it was that learners actually acquired when
they tried to learn an L2. This question was motivated by the recognition that
learners often failed initially to produce correct sentences and instead dis¬
played language that was markedly deviant from target language norms.
In order to answer this question, researchers collected samples of learner
language and tried to describe their main features. For example, the language
samples that learners produced were inspected for errors and these were then
classified. Alternatively, recordings of learners communicating with native
speakers or other learners were transcribed, specific grammatical features
16 Background

such as negatives or interrogatives were identified in the data, and descrip¬

tions of the ‘rules’ which could account for the learners’ productions were de¬
veloped. The aim of this research, then, was essentially descriptive—to
document what kind of language learners produced, to try to establish
whether it manifested regularities of some kind or other, and to find out how
it changed over time.

2 How do learners acquire a second language?

Not surprisingly, researchers were not content just to describe learner lan¬
guage; they also sought to explain it.That is, they wanted to account for why
learners made errors, why their language displayed marked regularities, and
why it changed systematically over time.
The answers were sought by formulating two further questions. The first
was ‘What contribution do external factors make to L2 acquisition?’ This in¬
volved considering the role played by the social situation in which learning
took place and how the language the learner was exposed to (i.e. the input)
accounted for acquisition as evident in the language the learner produced (i.e.
the output). Researchers seeking external explanations of learner language
made extensive use of ideas and methods from the sociolinguistic study of
language.
The second question was ‘What contribution do internal factors make to
L2 acquisition?’ In this case, explanations were sought in the mental
processes that the learner used to convert input into knowledge. Various pro¬
cesses have been identified. Some account for how the learner makes use of
existing knowledge (of the mother tongue, of general learning strategies, or of
the universal properties of language) to internalize knowledge of the L2.
These processes can be thought of as learning processes. They serve as the
means by which the learner constructs an interlanguage (a transitional system
reflecting the learner’s current L2 knowledge). Other processes account for
how the learner makes use of existing knowledge to cope with communica¬
tion difficulties. For example, sometimes the learner does not know the word
needed to communicate an idea clearly and has to resort to paraphrase or
word coinage. These processes are known as communication strategies.
Internal explanations have made extensive use of ideas drawn from both
cognitive psychology and linguistics.
A full explanation of L2 acquisition will need to take account of both ex¬
ternal and internal factors and how the two interrelate. We are still a long way
from such an explanation, however.
 Second language acquisition research: An overview 17

3 What differences are there in the way in which individual learners

acquire a second language?

In the case of questions (1) and (2) the focus is on learning. In question (3) the
focus is on the individual language learner. While much of the work that has
taken place in SLA research is based on the assumption that learner language
provides evidence of universal learning processes, there is also a long tradi¬
tion of research that has recognized that learners vary enormously in their
rate of learning, their approach to learning, and in their actual achievements.
The study of individual learner differences seeks to document the factors that
contribute to these kinds of variation.

4 What effects does instruction have on second language

acquisition?

Much of the early research investigated naturalistic L2 learners,3 motivated in

part by claims that classroom learning would proceed most smoothly if teach¬
ers stopped interfering in the learning process and left learners to learn in the
same way as children acquired their mother tongue (see Newmark 1966).
Increasingly, though, researchers have turned to studying the effects that in¬
struction of various kinds has on L2 acquisition. This research has been
motivated in part by a desire to address issues of general theoretical interest to
SLA research and also by a desire to improve the efficacy of language ped¬
agogy. The classroom affords an opportunity to control very precisely the
nature of the input that learners are exposed to. This in turn allows the
researcher to formulate and test very specific hypotheses regarding how par¬
ticular features of an L2 are acquired.
These four questions serve as a heuristic for exploring SLA research. They
have helped researchers to identify a number of key issues.

A framework for exploring second language acquisition

It is possible to identify a number of different areas of SLA that have been in¬
vestigated. These are shown in Table 1.1.
The first area of work concerns the description of the characteristics of
learner language. This provides the researcher with the main source of
information about how acquisition takes place. Four aspects of learner
language have received attention: (1) errors, (2) acquisition orders and
developmental sequences, (3) variability, and, more recently, (4) pragmatic
features relating to the way language is used in context for communicative
purposes. As we have already noted, one of the goals of SLA research is to
describe learner language and to show how it works as a system.
The second area concerns learner-external factors relating to the social
context of acquisition and to the input and interaction which the learner
experiences.
18 Background

The third area, learner-internal mechanisms, concerns how acquisition

takes place and how learners use their resources in communication. These
mechanisms are mental and largely hidden from view, although not necessar¬
ily completely unconscious. They relate to (1) the transfer of knowledge from
the learner’s LI, (2) the universal processes involved in converting input into
intake and restructuring existing L2 knowledge systems, (3) the utilization of
innate knowledge of linguistic universals, and (4) the processes for using L2
knowledge in performance, in particular those involved in dealing with prob¬
lems of communication. The study of learner-external factors and learner-
internal mechanisms constitutes an attempt to explain how L2 acquisition
takes place.
Finally, there is the question of individual learner differences and what
causes them. Learners set about the task of acquiring an L2 in different ways.
They differ with regard to such general factors as motivation and aptitude,
and also in the use of various strategies such as inferencing and self¬
monitoring for obtaining jriput and for learning from it.4 The study of these
general factors and learner strategies helps to explain why some learners learn
more rapidly than others and why they reach higher levels of proficiency.
It is not always clear where specific lines of enquiry fit best. For example,
the study of the communication processes involved in using L2 knowledge is
viewed by some researchers as an aspect of interaction (area 2), by others as
an aspect of mental processing (area 3) and by yet others as one type of learner
strategy (area 4). They have been included in area 3 in this book as a recogni¬
tion of the primacy of psycholinguistic accounts of communication processes
in recent research (see Bialystok 1990a).
Also, the four areas all interrelate, so not surprisingly many investigations
have covered more than one. For example, the errors that learners make (area

Focus on learning Focus on the

l
learner

Description Explanation

Area 1 Area 2 Area 3 Area 4

Characteristics of Learner-external Learner-internal The language learner
learner language factors mechanisms
errors social context LI transfer general factors e.g.
acquisition orders input and interaction learning processes motivation
and developmental communication learner strategies
sequences strategies
variability
knowledge of
pragmatic features linguistic universals

Table 1.1: A framework for investigating L2 acquisition

 Second language acquisition research: An overview 19

1) reflect the operation of internal processing mechanisms (area 3) and may

also be influenced by the social context in which learning takes place (area 2)
and the learner’s preferred learning style (area 4). One of the goals of this
book will be to explore the interactions between the four areas.

Learner language

The starting point of our exploration of L2 acquisition will be the study of

learner language. This area of SLA is central because it provides the data for
constructing and testing theories of L2 acquisition.

Errors

One of the first ways in which researchers tried to investigate L2 acquisition

was through the analysis of learner errors. The field of Error Analysis is
closely associated with the work of Corder, who published a number of sem¬
inal articles in the 1960s/1970s, (for example, Corder 1967 and 1971a) in
which he made out a case for examining errors as a way of investigating
learning processes. He also helped to develop a methodology for carrying out
an error analysis.
Much of the early work on learner errors focused on determining the extent
to which L2 acquisition was the result of LI transfer or of creative construc¬
tion (the construction of unique rules similar to those which children form in
the course of acquiring their mother tongue). The presence of errors that mir¬
rored LI structures was taken as evidence of transfer, while the presence of
errors similar to those observed in LI acquisition was indicative of creative
construction (i.e. intralingual). In a series of studies, Dulay and Burt analysed
errors produced by Spanish learners of English in order to determine which
errors were interlingual (indicative of transfer), and which ones were intra¬
lingual (indicative of creative construction). (See Dulay and Burt 1973,
1974a, and 1974b.)
The results of error analyses were used to refute behaviourist views of L2
learning, which were dominant at the time. According to these, L2 learning
took place in the same way as any other kind of learning—it involved proced¬
ures such as imitation, repetition, and reinforcement, which enabled learners
to develop ‘habits’ of the L2. The study of learner errors showed that al¬
though many errors were caused by transferring LI ‘habits’, many more were
not; learners often contributed creatively to the process of learning. They also
indicated that learners appeared to go through stages of acquisition, as the
nature of the errors they made varied according to their level of development.
Error analysis, however, as practised in the sixties and seventies, was an im¬
perfect research tool. It could not show when learners resorted to avoidance
(Schachter 1974) and, as it ignored what learners could do correctly, it only
looked at part of learner language. Also, the methodology of error analysis
20 Background

was suspect in a number of respects. For example, it was not entirely clear
what constituted an ‘error’ and it proved difficult to prepare rigorous descrip¬
tions of errors. As a result, many studies were unreliable and difficult to rep¬
licate. It is not surprising, perhaps, that error analysis has fallen out of favour
with many researchers. However, the study of learner errors can still serve as
a useful tool and is still undertaken, often as a means of investigating a specific
research question. For example, Bardovi-Harlig and Bofman (1989) em¬
ployed error analysis techniques to investigate the linguistic differences be¬
tween advanced learners who were successful on a university placement test
and those who were unsuccessful. Also, error analysis can prove illuminative
when employed in the detailed, qualitative analysis of learner language (see
Taylor 1986).
Work on errors in learner language is discussed in Chapter 2.

Acquisition orders and developmental sequences

The language that learners produce provides evidence that they acquire dif¬
ferent morphological features in a fixed order and also that they pass through
a sequence of developmental stages in the acquisition of specific syntactical
features. Early evidence for this came from studies of learners’ errors, but
more convincing evidence was provided by a series of cross-sectional and
longitudinal studies carried out in the seventies and eighties.
There is also evidence to support the existence of a general pattern of devel¬
opment. This general pattern may begin with a silent period, particularly in
children. Many of the initial utterances that learners produce are for¬
mulas—ready-made chunks like ‘I don’t know’ and ‘What’s this?’ Gradually
learners begin to produce more creative utterances, although to begin with
these are often propositionally reduced and morphologically simple (for ex¬
ample, ‘Kick’ for ‘He kicked me’). It has been suggested that learners work on
the formulas they have learnt by substituting elements in them and joining
them together to form complex structures (Wong-Fillmore 1976; Ellis
1984a). Slowly learners master the basic syntax of the L2, learning the major
permutations in word order found in the target language. At this stage some
morphological features are also acquired, but many do not appear until much
later, if at all.
Evidence for a definite order in the acquisition of specific grammatical fea¬
tures was provided by the morpheme studies. These investigated a set of
grammatical functors in English (for example, plural -s, past tense -ed, verb
-ing, articles) and claimed to show that there was a natural order of acquisi¬
tion (see Krashen 1977 for a review of these studies) that was not influenced
by the learner’s age or first language. Major variation in the order was only
found when the learner had the opportunity to monitor output (i.e. make use
of conscious or ‘learnt’ knowledge of grammatical rules). Most of these
studies were cross-sectional in design. The claims they made about an order
 Second language acquisition research: An overview 21

of acquisition were based on the rank order of accuracy with which learners
performed the different functors in output collected at a single point in time.
The assumption was that learners must have acquired those features they per¬
formed more accurately before those they performed less accurately. The
morpheme studies have been criticized on the grounds that this assumption is
not justified (see Hatch 1978d and 1983a). Other problems have also been
identified.
Longitudinal studies of learner language afforded evidence of develop¬
mental sequences. Early work, based on case studies of individual learners
(for example, Ravem 1968; Huang 1970; Wode 1976 and 1978) demon¬
strated that learners with different language backgrounds followed a remark¬
ably similar path of development when trying to produce structures such as
English negatives and interrogatives. The learners appeared to construct a
series of transitional rules before they mastered the target language rules. Fur¬
thermore, the developmental route they followed seemed to closely follow
that reported for the acquisition of the same structures by children learning
English as their first language.
Some of the most powerful evidence for developmental sequences has come
from studies of the L2 acquisition of German, such as the Zweitspracherwerb
Italienischer, Spanischer und Portugiesischer Arbeiter (ZISA) project (Meisel,
Clahsen, and Pienemann 1981). This project, which studied adult and child
learners with romance language backgrounds, found that some grammatical
features of German were not acquired in any definite sequence, but that
others were. Word order rules were clearly developmental. Learners—chil¬
dren and adults—passed through a well-defined sequence. Other features
such as copula ‘sein’ were subject to considerable inter-learner variation. Sub¬
sequent work in Australia (for example, Johnston 1985) has demonstrated a
similar sequence for word order features of English and similar variation in
features such as copula ‘be’.
The existence of developmental sequences is one of the most important
findings of SLA research to date. There is now general acceptance in the SLA
research community that the acquisition of an L2 grammar, like the acquisi¬
tion of an LI grammar, occurs in stages. However, it should be noted that
although general developmental sequences have been attested in learners in
different situations and with differing backgrounds, variations in the specific
order in which particular features occur have also been found. Lightbown
(1984) claims that for every study that supports the idea of a universal se¬
quence, there is another study that provides counter-evidence.
An important issue—for both theory building and for language
pedagogy—concerns the effect that formal instruction has on the acquisition
of grammatical features. A number of studies have investigated whether
teaching specific grammatical features results in their acquisition (Pienemann
1984; White, Spada, Lightbown, and Ranta 1991), while others have invest¬
igated the effects of formal instruction by comparing the language produced
22 Background

by naturalistic and instructed learners (for example, Pica 1983; Ellis 1984c).
The results of these studies are not easy to interpret, partly because the con¬
cept of what ‘instruction’ means has varied greatly (see Ellis 1990a) and
partly because researchers have worked with very different operational def¬
initions of acquisition. However, there is growing evidence to indicate that
grammar instruction does work, providing learners are ready to assimilate
the new target rule into their mental grammars, although instruction does not
appear to enable learners to ‘beat’ a developmental sequence.
The evidence for a general pattern of development, acquisition orders and
developmental sequences is examined in Chapter 3. The effect of formal
instruction on the learning of grammatical features, including those that are
developmental, is considered in Chapter 14.

Variability

Learner language, like the language of native speakers, appears to be inher¬

ently variable. Learners frequently use one structure on one occasion and a
different structure on another. Teachers are very familiar with learners who
use a feature such as 3rd person -s correctly at one time only to omit it at an¬
other. Researchers who have investigated learner language frequently com¬
ment on the ‘constant development and concomitant variation’ (Cancino,
Rosansky, and Schumann 1978: 209) in their subjects’ use of the L2. This
variation is, of course, not confined to learners, as native speakers have also
been shown to vary in their choice of linguistic features (Labov 1970). How¬
ever, language learners seem particularly prone to variability.
A key issue is the extent to which this variability is systematic. Much of it
undoubtedly is. Learners alternate their use of linguistic forms according to
linguistic context. For example, they are less likely to omit copula ‘be’ if it is
preceded by a pronoun subject than if it is preceded by a noun subject (Ellis
1988a). Learners also vary according to situational context, in particular ac¬
cording to who they are speaking to (Beebe 1980). A similar stylistic con¬
tinuum has been found in learner language as in native speaker speech (see
Tarone 1982). Learners are more likely to use correct target language forms
in situations and tasks that call for a careful style (i.e. formal language use)
and more likely to use transitional, learner forms in their vernacular style (i.e.
in informal, everyday language use). Variability is also systematic in another
way. Learners may make deliberate use of whatever linguistic knowledge
they possess in order to distinguish different functional or semantic meanings
that are important to them. Again, in this they are no different from native
speakers. However, they often create unique form-function correspondences
of a kind not found in the target language. Lor example, at a certain stage of
development learners may decide to use one negative rule for making state¬
ments (for example, ‘No making noise’ = ‘I’m not making a noise’) and an¬
other for commands (for example, ‘Don’t make a noise’).
 Second language acquisition research: An overview 23

There is less agreement as to whether some of the variability in learner lan¬

guage is non-systematic. In Ellis 1985c and 1989c, I argue that when learners
acquire new forms they are likely to first use these in free variation with
existing forms. Later they will either drop an ‘old’ form or use the two forms
systematically in some way. Other researchers (for example, Schachter
1986a) consider non-systematic variability of little significance, however.
There is also no agreement as to how important variability is for under¬
standing learning processes. Whereas Tarone and I see it as providing crucial
evidence about the ways learners set about constructing an L2 grammar,
Gregg (1990) argues that it is an uninteresting and trivial phenomenon
because it only reflects performance and sheds no light on the learner’s
competence. The nature of variability in learner language is explored in
Chapter 4, while the major theoretical positions are outlined in Chapter 9.

Pragmatic features

Pragmatics is the study of how language is used in communication. It covers a

wide range of phenomena including deixis (i.e. the ways in which language
encodes features of the context of utterance), conversational implicature and
presupposition (i.e. the way language is used to convey meanings that are not
actually encoded linguistically), illocutionary acts (i.e. the use of language to
perform speech acts such as stating, questioning, and directing), conversa¬
tional structure (i.e. the way in which conversations are organized across
turns), and repair (i.e. the conversational work undertaken to deal with mis-
communications of various kinds). Pragmatics is particularly concerned with
appropriateness, both with regard to what is said in a particular context and
how it is said.
Early work in SLA research did not entirely neglect pragmatic aspects of
learner language. However, they were attended to only in so far as they
helped to explain how learners acquired grammatical competence. Hatch
(1978b), for instance, examined the ways in which learners develop topics as
a way of exploring why certain grammatical features are acquired before
others. Only recently, however, have researchers considered the acquisition
of pragmatic competence in its own right. The bulk of the work to date has fo¬
cused on specific illocutionary acts. Kasper and Dahl (1991) provide a useful
overview of work on interlanguage pragmatics.
Learners have to learn when it is appropriate to perform a particular lan¬
guage function and also how to encode it. They frequently experience prob¬
lems with both. Thomas (1983) distinguishes sociopragmatic failure, which
occurs when learners produce socially inappropriate behaviour and prag-
malinguistic failure, which occurs when learners do not express themselves in
a linguistically appropriate manner. A number of studies have demonstrated
the difficulties that learners experience. Beebe and Takahashi (1989a), for in¬
stance, report on the way in which Japanese and native speakers of English
24 Background

differed in their performance of disagreements and giving embarrassing

information. Blum-Kulka, House and Kasper (1989a) report on the diffi¬
culties that even advanced L2 learners have in encoding requests and
apologies in socially appropriate ways.
Chapter 5 looks at the research which has investigated how speech acts are
encoded.

Learner-external factors
External factors relate to the environment in which learning takes place.
What exactly their role in acquisition is and how important they are remain
controversial issues. Behaviourist theories of learning view external factors as
of central importance. Mentalist theories, however, emphasize the role
played by learner internal factors, crediting learners with a Language Ac¬
quisition Device that enables them to work on what they hear and to extract
the abstract ‘rules’ that account for how the language is organized. Cognitive
theories of language acquisition tend to be interactional in the sense that they
emphasize the joint contribution of external and internal factors.

Social factors
Social factors probably have an indirect rather than a direct effect on L2
learning. In particular, they are likely to be mediated by the attitudes that the
learners hold. Social factors shape learners’ attitudes, which, in turn, deter¬
mine learning outcomes.
The impact of social factors on learning outcomes has been studied in rela¬
tion to L2 proficiency rather than developmental patterns, as it has been gen¬
erally assumed that social factors do not directly influence the process of L2
acquisition. There are alternative models of L2 proficiency (see Bachman
1990) but two have figured strongly in SLA research. Cummins (1983) distin¬
guishes basic interpersonal communication skills (BICS) from cognitive aca¬
demic language proficiency (CALP). Canale and Swain (1980) distinguish
four ‘modules’ of ability relating to different types of language know¬
ledge—grammar, sociolinguistic, discourse and strategic. The principal ques¬
tion that has been addressed is ‘To what extent and in what ways do social
factors affect the L2 proficiency attained by different groups of learners?’
The social factors that influence L2 acquisition are likely to differ accord¬
ing to social context. A key distinction is that between a natural and an educa¬
tional setting. A general assumption is that the learning that takes place in
these two types of setting is very different (see d’Anglejan 1978), but the
extent to which this is the case may well depend on the nature of the more
specific settings in which learners find themselves. Natural settings can be
distinguished according to whether the L2 serves as a native language for the
majority (for example, learners of L2 English in the United States), whether it
 Second language acquisition research: An overview 25

serves as an official language when the majority speaks some other language
(for example, L2 English in Nigeria or India), or whether it is used by lin¬
guistically heterogeneous groups in international settings (for example, the
use of L2 English during a business meeting in Japan). Educational settings
can be distinguished according to whether they involve segregation (i.e. the
learners are taught the L2 or taught through the medium of the L2 separately
from the majority group), mother tongue maintenance (i.e. an attempt is
made to ensure that a minority group’s LI is taught and used in the educa¬
tional setting), submersion (i.e. the L2 learner is taught in classes where LI
speakers are dominant), immersion (i.e. learners with a high-status LI are
taught through the medium of the L2 in classes containing only such learners,
usually by bilingual teachers), or foreign language classrooms (for example,
English language classrooms in Japan). Different types and levels of L2 profi¬
ciency are typically associated with each type of setting, although, of course,
there can be considerable variation among learners within each setting.
One way in which the social context affects learning outcomes in these dif¬
ferent settings is by influencing the learners’ choice of reference group. Beebe
(1985: 404) has argued that learners are ‘active participants in choosing the
target language models they prefer’ in majority language settings. She distin¬
guishes ‘unmarked’ and ‘marked’ choices depending on whether the choice is
the expected one given the learners’ social position or an unexpected one. In
official language settings, the preferred model is often some indigenized vari¬
ety of the L2, while in foreign language settings it is likely to be a standard nat¬
ive-speaker variety.
There has been relatively little research which has investigated the effects of
particular social factors on L2 proficiency. However, a few studies relating to
key sociolinguistic variables such as age, sex, social class, and ethnic member¬
ship have been carried out. The general findings are that younger learners do
better than older learners (although this may also reflect psycholinguistic fac¬
tors), females outperform males, middle class learners achieve higher levels of
academic language proficiency than working class learners, and learners from
an ethnic group that is culturally similar to the target language group tend to
be more successful than learners from an ethnic group that is culturally dis¬
tant. There are some interesting exceptions to these unsurprising findings,
however. Lor example, working class children do as well as middle class chil¬
dren in immersion settings (Holobrow, Genesee, and Lambert 1991).
A number of theories have been developed to explain the relationship be¬
tween social factors and L2 acquisition. Schumann (1978a) suggests that
learners vary in the extent to which they acculturate (i.e. adapt to the target
language culture). Some learners remain at a social distance from the target
language community and as a result tend to pidginize (i.e. develop only a very
basic competence in the L2). Other learners assimilate and develop a high
level of proficiency in the L2. According to Schumann’s model, social factors
determine how much contact individual learners have with the L2. Giles and
26 Background

Byrne (1982) have developed a similar theory to account for the effect of vari¬
ous social factors on L2 acquisition. They argue that the learner must be pre¬
pared to ‘converge’ towards the norms of the target language and that this
only takes place if certain positive social factors come into play. For example,
learners are more likely to converge if they perceive their culture to be equal
or superior in status to that of target language speakers. This inter-group
model of L2 acquisition emphasizes the role that social factors play in de¬
termining the kinds of interactions learners participate in. According to
Gardner’s Socio-educational Model of L2 Learning (Gardner 1985), the so¬
cial and cultural milieu in which learners grow up determines the attitudes
and motivational orientation that learners hold towards the target language,
its speakers, and its culture—which in turn influence the sorts of learning
behaviours learners engage in, and thereby learning outcomes.
The relationship between setting and L2 learning, the learner’s choice of
preference model, the impact of individual social factors such as social class
and ethnic membership on learning and the different social theories of L2
acquisition are examined in Chapter 6.

Input and interaction

It is self-evident that L2 acquisition can only take place when the learner has
access to input in the L2. This input may come in written or spoken form. In
the case of spoken input, it may occur in the context of interaction (i.e. the
learner’s attempts to converse with a native speaker, a teacher, or another
learner) or in the context of non-reciprocal discourse (for example, listening
to the radio or watching a film). The study of input and interaction has in¬
volved the description of the adjustments which are found in language ad¬
dressed to learners (i.e. foreigner talk and teacher talk) and also the analysis
of discourse involving L2 learners.
There is little agreement about the role that input plays in L2 acquisition.
Behaviourist theories emphasize its importance, claiming that the whole pro¬
cess of acquisition can be controlled by presenting learners with input in the
right-sized doses and then reinforcing their attempts to practise them. Ac¬
cording to this view of learning there is little room for any active processing
by the learner. In the 1960s, however, behaviourist accounts of learning were
challenged, most notably by Chomsky. It was pointed out that in many cases
there was a very poor match between the kind of language found in the input
learners received and the kind of language they themselves produced. It was
argued that this could be best explained by hypothesizing a set of mental pro¬
cesses which took place inside the mind of the learner and which converted
the language in the input into a form that the learner could store and handle in
production. This mentalist view of input has itself been challenged by re¬
searchers on a number of grounds. For example, it has been shown that inter¬
action can provide learners with ‘scaffolding’ that enables them to produce
 Second language acquisition research: An overview 27

structures that would be beyond them, if left to their own resources. Re¬
searchers who emphasize the importance of input and interaction suggest
that learners acquire a language through the process of learning how to com¬
municate in it (see Hatch 1978b). There is currently little support for behavi¬
ourist views of language acquisition, but the debate continues between those
who argue that input serves only as a trigger that sets off some internal lan¬
guage acquisition device (for example, White 1987a) and those who argue
that input shaped through interaction contributes directly and powerfully to
acquisition (for example, Long 1981a).
The relationship between input and acquisition has been explored in a
number of ways. Several studies (for example, Snow and Hoefnagel-Hohle
1982) have examined the relationship between the frequency with which dif¬
ferent linguistic items occur in the input and the order of acquisition of the
same features. These studies have produced mixed results and are of limited
value as they are all correlational in nature, thus making it difficult to make
statements about cause and effect. More important are studies which have in¬
vestigated the role of comprehensible input. Krashen (1985) has proposed the
Input Hypothesis, according to which learners acquire morphological fea¬
tures in a natural order as a result of comprehending input addressed to them.
Long (1981a) has argued that input which is made comprehensible by means
of the conversational adjustments that occur when there is a comprehension
problem is especially important for acquisition. Krashen’s and Long’s pro¬
posals have led to several studies that have investigated what factors are in¬
volved in making input comprehensible (for example, Parker and Chaudron
1987; Pica, Young, and Doughty 1987). However, relatively few studies to
date have attempted to show that comprehensible input actually leads to the
acquisition of new linguistic features. Also, there are theoretical objections to
the position adopted by Krashen and Long; it has been pointed out, for ex¬
ample, that the processes of comprehension and of acquisition are not the
same, nor are they necessarily related (Sharwood Smith 1986).
Other researchers have emphasized the role of learner output in promoting
acquisition. Swain (1985), for instance, has put forward the comprehensible
output hypothesis, which states that learners need opportunities for ‘pushed
output’ (i.e. speech or writing that makes demands on them for correct and
appropriate use of the L2) in order to develop certain grammatical features
that do not appear to be acquired purely on the basis of comprehending input.
Swain’s case rests on the different psycholinguistic requirements of compre¬
hension and production; whereas successful comprehension is possible with¬
out a full linguistic analysis of the input, correct production requires learners
to construct sentence plans for their messages. Again, however, there is little
direct evidence to support Swain’s hypothesis. Finally, other researchers (for
example, Hatch 1978b) have tried to show how the process of constructing
discourse collaboratively with an interlocutor helps learners to develop new
syntactic structures.
28 Background

There has also been considerable interest in classroom interaction on the

grounds that the opportunities for learning which it affords constitute the
major determinant of acquisition in this setting (Allwright 1984). Much of
the research has been based on Krashen’s and Long’s views about the import¬
ance of comprehensible input. In particular, researchers have sought to estab¬
lish what classroom conditions result in the kinds of conversational
adjustments that Long hypothesizes are important for acquisition. To this
end, they have compared the interaction that takes place in small-group work
with that occurring in teacher-led lessons (for example, Pica and Doughty
1985a and 1985b). They have also investigated the effect of task design on
interaction. Certain kinds of tasks appear to result in more modified inter¬
action than others. For example, tasks that involve a two-way exchange of
information—such as when learners each hold part of the information which
they need to solve a problem jointly—have been shown promote extensive
meaning negotiation.
The role of input in L2 acquisition has proved to be a controversial issue. It
is probably true to say that there has been too much theorizing and not
enough empirical research. Very few studies have actually sought to demon¬
strate if and how input affects acquisition. Much of the research has been de¬
scriptive in nature—describing the kinds of linguistic modifications that
occur in foreigner talk, or the devices used to negotiate meaning, or the inter¬
actions that result from asking learners to perform tasks in different
conditions.
The role of input and interaction is considered in Chapter 7. Chapter 13
focuses on interaction in the L2 classroom.

Learner-internal factors

Any theory of SLA needs to provide an account of learner-internal factors.

These factors are, of course, not directly observable. They are covert and can
only be inferred by studying learner output and, to some extent, learners’ re¬
ports of how they learn.

Language transfer

LI transfer usually refers to the incorporation of features of the LI into the

knowledge systems of the L2 which the learner is trying to build. It is import¬
ant to distinguish this learning process from other processes which involve
the use of the LI for purposes of communication (see below). Both translation
and borrowing (Corder 1983)—the use of the LI to deal with some commun¬
ication problem—are examples of communication transfer, as are code¬
mixing (the use of both the LI and L2 in the construction of the same sen¬
tence) and code-switching (the alternative use of the LI and L2 within a
 Second language acquisition research: An overview 29

discourse). It is, of course, not easy to distinguish empirically the process of

transfer in L2 learning and the use of the LI as a communication process.
Views about language transfer have undergone considerable change. Ini¬
tially, transfer was understood within a behaviourist framework of learning.
It was assumed that the ‘habits’ of the LI would be carried over into the L2. In
cases where the target language differed from the LI this would result in inter¬
ference or rjegative transfer. In cases where the patterns of the LI and the tar¬
get language were similar, positive transfer would occur. Thus, the LI could
both impede and facilitate the acquisition of the L2. In order to try to predict
when interference would take place linguists carried out elaborate contrastive
analyses of the native and target languages. It was not until the late 1960s,
however, that the claims of the contrastive analysis hypothesis were investig¬
ated empirically. The results of error analysis studies (for example, Dulay and
Burt 1974a) cast doubts on the validity as these claims. Transfer often did not
take place when there were differences between the target and native lan¬
guages. Also, many of the errors that learners made appeared to reflect intra-
lingual processes (i.e. they were the result of processes based on the learner’s
existing knowledge of the L2) rather than interference. This led to the ad¬
vancement of a minimalist position regarding LI influence (see Dulay and
Burt 1974a).
More recently the importance of transfer has once again been acknow¬
ledged. Odlin (1989), reflecting a consensus that grew throughout the 1980s,
comments:

Despite the counterarguments ... there is a large and growing body of re¬
search that indicates that transfer is indeed a very important factor in
second language acquisition (1989:4).

Evidence for transfer in all aspects of language—phonology, syntax, semant¬

ics, and pragmatics—is now abundant. Furthermore, there is recognition of
the fact that transfer may not always manifest itself as errors (the focus of
early studies), but also as avoidance, overuse, and facilitation. It is now gener¬
ally acknowledged that transfer works in complex ways and that it consti¬
tutes only one of several processes involved in L2 acquisition (see Zobl
1980a). Much effort has gone into identifying the conditions under which
transfer is likely to occur. Various factors have been identified which influ¬
ence transfer. For example, Kellerman (1978) has shown that learners’ per¬
ception of the distance between their native language and the target language
(i.e. their psychotypology) affects whether they transfer or not. Other studies
(for example, Wode 1976) have shown that transfer may be a developmental
phenomenon in that it occurs only when the learner reaches a ‘natural’ stage
of acquisition which bears a crucial similarity to some native language struc¬
ture. Transfer may also be affected by markedness. According to one defini¬
tion of markedness, a marked linguistic structure is one that can be used with
fewer constraints than a related unmarked one. For example, the adjective
30 Background

‘old’ is considered less marked than ‘young’ because it can be used in both
questions (for example, ‘How old are you?’) and statements (for example,
‘He is very old.’), whereas ‘young’ cannot be similarly used in questions (for
example, *‘How young are you?’). Learners seem more likely to transfer
unmarked native language features than marked ones, particularly if the cor¬
responding feature in their target language is marked (Zobl 1984). These and
other factors determine whether transfer takes place.
Transfer as a learning process and as an aid to communication is con¬
sidered in Chapter 8.

Cognitive accounts of second language acquisition

We have just noted that learners may use their LI as both a means of learning
an L2 (when we speak of transfer) and as a means of communicating. This is
an example of a broad distinction between learning processes and commun¬
ication processes. The former consist of the mechanisms that learners use to
(1) notice features in the input, (2) compare these features with those that are
currently part of their mental grammars or interlanguages, and (3) integrate
the new features into their interlanguages. The latter consist of conscious or
potentially conscious attempts on the part of the learner to employ their avail¬
able linguistic resources, for example by developing their ability to use the L2
fluently or by compensating for inadequate knowledge when communicating
a particular message.
In addition to transfer, learners use a variety of processes to learn an L2. A
number of different theories of SLA have provided accounts of them. These
theories are cognitive in nature, i.e. they attempt to account for the mental
processes that enable learners to work on input, and for the knowledge-
systems which they construct and manifest in output.
Perhaps the best known is interlanguage theory. The term interlanguage
was coined by Selinker (1972) to refer to the interim grammars which learn¬
ers build on their way to full target language competence. As McLaughlin.
(1987) observes, interlanguage theory has undergone almost constant devel¬
opment, but one common theme is the notion of hypothesis testing, i.e. the
idea that learners form hypotheses about what the rules of the target language
are and then set about testing them, confirming them if they find supportive
evidence in the input and rejecting them if they receive negative evidence. This
process takes place largely on a subconscious level. Interlanguage theory has
also identified a number of other, more specific processes such as overgeneral¬
ization (i.e. the extension of an L2 rule to a context in which it does not apply
in the target language) and simplification (i.e. the reduction of the target lan¬
guage system to a simpler form). Early work on learner errors, acquisition
orders, and developmental sequences was closely related to interlanguage
theory.
 Second language acquisition research: An overview 31

Parallel to interlanguage theory, and influenced by it, are a number of

theories which seek to account for the observed characteristics of learner
language, and in particular for developmental patterns. Several theories rest
on the distinction between implicit (i.e. intuitive and unconscious) and expli¬
cit (i.e. metalingual and conscious) L2 knowledge, differing in whether these
are seen as entirely distinct and unconnected, as in Krashen’s Monitor Theory
(Krashen 1981), or whether they are hypothesized to interface, as in
Bialystok’s Model of Second Language Learning (Bialystok 1978). More
recent discussion of the implicit/explicit distinction has focused on the role of
consciousness in L2 learning (see Schmidt 1990).
Other theories have been based on research that shows learner language to
be highly variable. Tarone (1979, 1982, 1983) has argued that learners con¬
struct variable interlanguage grammars consisting of different styles. The
learners’ vernacular style develops both as a result of directly processing in¬
put, and as a result of the spread of forms from their careful style. In Ellis
1985c and 1989a, I suggest that interlanguage development entails three con¬
current phases, involving: (1) innovation (i.e. the acquisition of new forms),
(2) elaboration (i.e. the complexification that takes place as the learner dis¬
covers the contextual uses of a form), and (3) revision (i.e. the adjustments
that are made to the entire system as a result of innovation and elaboration).
Closely related to variationist accounts of interlanguage are functionalist
models (see Tomlin 1990; Dittmar 1992). These suggest that variability is a
reflex of different modes of language use. Givon (1979), for instance, suggests
that L2 development occurs as learners move from processing language in
pragmatic terms (for example, as topic-comment structures) to processing it
in grammatical terms (for example, as subject-predicate structures). Func¬
tionalist accounts also emphasize the importance of the form-function net¬
works that learners are believed to construct. According to the Competition
Model (Bates and MacWhinney 1982), language acquisition involves the
learner attending to both form and function. There are inherent human ten¬
dencies that induce learners to perceive and process some forms before others
but also a form is more likely to be acquired if it realizes a meaning that is
transparent and important to the learner. According to this model, learners
assign ‘weights’ to different form-function mappings and the process of ac¬
quisition involves adjusting these weights until they match those found in the
input. For example, the learner of F2 English may initially work on the as¬
sumption that the semantic function ‘agent’ is invariably designated by word
order (i.e. any noun occurring before a verb is an agent). This may lead the
learner to misconstrue sentences like ‘The dog was chased by the cat’. Sub¬
sequently, the ‘weight’ connecting word order to agency will be weakened as
the learner recognizes the validity of other formal cues in the input (for ex¬
ample, the presence of ‘by + noun’ after the verb) and establishes new form-
function mappings which compete with the original.
32 Background

In another approach, researchers have sought to identify the operating

principles which learners use to convert input into a form which they can
store. Slobin (1973, 1985b) has identified a number of operating principles
which make up the ‘Language Making Capacity’ of children acquiring their
LI. Examples are ‘pay attention to the ends of words’ and ‘avoid interruption
and rearrangement of linguistic units’. These principles have shown to be pre¬
sent in the LI acquisition of very different languages. Andersen (1984a;
1990) has built on Slobin’s work and identified a number of macro-principles
found in L2 acquisition. The one-to-one principle, for example, states that ‘an
interlanguage system should be constructed in such a way that the intended
underlying meaning is expressed with one clear invariant surface form or con¬
struction’ (Andersen 1984a: 79).
An adequate theory of L2 acquisition needs to account for why some lin¬
guistic features are acquired before others and why learners manifest devel¬
opmental sequences when acquiring specific features. Perhaps the theory that
comes closest to satisfying this requirement is Johnston and Pienemann’s
(1986) model of L2 acquisition. This model is a development of the
Multidimensional Model advanced by Meisel, Clahsen, and Pienemann
(1981). It accounts for the developmental properties of interlanguage in
terms of a number of processing operations, which are organized hierarch¬
ically according to the order in which they are evident in learners’ production.
Thus, initially learners rely on non-linguistic processing devices (for example,
formulas and lexical items that are not assigned to grammatical categories)
and then move through a series of stages until they are able to carry out more
complex grammatical operations. This model also recognizes and provides
an explanation for individual learner variation.
Other cognitive models have attempted to explain L2 acquisition in terms
of a general theory of skill learning. Andersen’s Adaptive Control of Thought
(ACT) Model sees language acquisition, like all other kinds of learning, as a
process of proceduralizing ‘declarative knowledge’ (i.e. knowledge stored as
facts). This takes place through ‘practice’. McLaughlin (1987) proposes a
Cognitive Theory based on information processing. Learners routinize lin¬
guistic information that is initially only available for use through controlled
processing and this frees them to attend to new information in the input. Also,
learners restructure their rule-based representations of the L2 at certain crit¬
ical points in the learning process. As in the case of Andersen, McLaughlin
sets considerable store on ‘practice’ as the mechanism of change.
All these theories are based on the assumption that learners form mental
representations of ‘rules’ and that these rules guide the learner in using the L2
in performance. A ‘rule’ consists of some form of abstract generalization of a
linguistic property—the fact that past tense forms in English consist of a verb
stem + -ed, for example. However, another approach to explaining language
acquisition dispenses altogether with the idea of rules. According to a Parallel
Distributed Processing Model (Rumelhart and McClelland 1986;
 Second language acquisition research: An overview 33

Rumelhart, McClelland, and the PDP Research Group 1986), knowledge

takes the form of a network of interconnections between ‘units’ that do not
correspond to any particular linguistic construct.’ Learning consists of the
modification of the strengths of these connections as a response to input stim¬
uli and is complete when the network corresponds to that found in native
speakers. Researchers construct mathematical models of the kinds of net¬
works that they think can account for learning and then test them in com¬
puter simulations. The criterion of success is whether the model produces
similar stages of development to those that have been observed in natural
language acquisition. As yet, however, there have been few PDP models
specifically designed to examine L2 acquisition (see Gasser 1990).
In addition to acquiring knowledge of the L2, learners also need to develop
control over existing knowledge, as suggested by Anderson’s account of pro-
ceduralization and McLaughlin’s views on controlled/automatic processing.
The acquisition of ‘control’ is specifically considered in work on L2 speech
planning by researchers at the University of Kassel in Germany (see Dechert,
Mohle, and Raupach 1984) and also in the extensive work on learners’ com¬
munication strategies (see Bialystok 1990a for an overview). Speech planning
has been considered in terms of temporal variables, such as speech rate and
pause length, and hesitation phenomena. A number of frameworks for clas¬
sifying communication strategies have been developed. Fserch and Kasper
(1980) propose a basic distinction between reduction strategies (i.e. strategies
that attempt to do away with the communication problem) and achievement
strategies (i.e. strategies designed to achieve the original goal). Poulisse
(1990a) proposes a distinction between conceptual strategies (i.e. strategies
that attempt to get round a problem by referring to a concept obliquely in
some way) and linguistic strategies (i.e. an attempt to get round a problem by
exploiting LI or L2 linguistic knowledge). Much of the work on communica¬
tion strategies has focused on trying to identify instances of their use and on
developing a reliable typology, although there is now some work that has ex¬
amined their comparative effectiveness in facilitating communication. There
is still very little known about how the use of communication strategies con¬
tributes to L2 learning. One interesting possibility is that the acquisition of
knowledge and of control occur independently, with learners differing in the
emphasis they give to each aspect.
Learning and communication processes are examined in detail in
Chapter 9.

Linguistic universal and second language acquisition

Mentalist theories of L2 acquisition emphasize the role of innate knowledge.
This takes the form of a language acquisition device which helps the learner to
discover the rules of the target language grammar. This device contains
knowledge of linguistic universal.
34 Background

There have been two approaches to the study of the role of linguistic uni¬
versal in L2 acquisition. A number of linguists have set about identifying typ¬
ological universals through the study of a large number of languages drawn
from different language families. They have then gone on to advance a num¬
ber of possible explanations (including mentalist ones) for the existence of
these universals. Other linguists—those belonging to the generative school
associated with Chomsky—have studied individual languages in great depth
in order to identify the principles of grammar which underlie and govern spe¬
cific rules (i.e. Universal Grammar). For example, the Phrase Structure Prin¬
ciple states that all languages are made up of phrases consisting of a head and
a complement (see Cook 1988 for an account of this and other principles).
Thus, in the English prepositional phrase ‘in the classroom’ ‘in’ constitutes
the head and ‘the classroom’, the complement. Principles are often parame¬
terized—that is, they allow for various options. Thus in the case of the Phrase
Structure Principle, one option is that the head precedes the complement (as
in English), while another is that the head follows the complement (as in
Japanese). Principles such as the Phrase Structure Principle can govern
clusters of features. Thus, if a language manifests head initial verb phrases
(for example, verb + direct object) it is also likely to manifest prepositions,
while if a language has head final verb phrases (for example, direct object +
verb) it will probably also have postpositions. In this way, the presence of one
feature in a language implies the presence of other, related features.
One of the key arguments advanced by those claiming a role for linguistic
universals is that of the logical problem of acquisition (see White 1989a). Ac¬
cording to this, the input to which the learner is exposed underdetermines lin¬
guistic competence, i.e. learners are unable to discover some of the rules of the
target language purely on the basis of input because it does not supply them
with all the information they need. It follows that they must rely on other
sources of information. These sources are knowledge of linguistic universals
and in the case of L2 learners, knowledge of their LI.
A second key argument concerns the ‘unlearning’ of a wrongly formulated
rule. It is claimed that in certain cases unlearning can only take place if the
learner is supplied with negative feedback in the form of overt corrections,
such as those that occur in many language classrooms. In the case of LI ac¬
quisition such feedback is rare. It is argued that children are prevented from
forming rules which they cannot unlearn without negative feedback by their
knowledge of linguistic universals. It is not clear, however, whether this argu¬
ment applies to L2 acquisition. For a start, L2 learners may be more prepared
to abandon previously learnt rules than LI learners (Rutherford 1989). Also,
it is not yet clear whether L2 learners have continued access to Universal
Grammar, or whether they rely on other methods of learning that involve
general learning processes and access to negative feedback (see Bley-Vroman
1989) or, of course, whether they rely on both.
 Second language acquisition research: An overview 35

Knowledge of linguistic universal may help to shape L2 acquisition in a

number of ways. First, it can provide an explanation for developmental se¬
quences. Learners may acquire rules that reflect universal principles before
they acquire rules that do not. Or they may opt for parameters that are less
marked before those that are more marked.6 Second, knowledge of linguistic
universal may enable the learner to go beyond the input. This can happen in
the case of clusters of features where an implicational relationship holds.
Thus, once learners of L2 Japanese have discovered that the verb comes at the
end of the clause in Japanese, they may be able to infer that it also has post¬
positions rather than prepositions. Knowledge of linguistic universal may
provide the learner with a projection capacity (Zobl 1983b). Third, know¬
ledge of linguistic universal may also be involved in transfer. Learners may
be more prepared to transfer LI features if these conform to universal
principles.
The study of linguistic universal in L2 acquisition has attracted consider¬
able attention in recent years. It has served as one of the ways in which the in¬
terests of linguists and SLA researchers can be brought together. Chapter 10
considers the role of linguistic universal in L2 acquisition.

The language learner

Whereas the bulk of SLA research has addressed the universal characteristics
of L2 acquisition, there is also a growing body of research on individual learn¬
er differences (see Skehan 1989). It is convenient to draw a distinction
between research that has investigated how general factors, such as motiva¬
tion, contribute to individual learner differences and research based on the
strategies that learners deliberately employ to enhance their learning of the
L2.

General factors and individual differences

Learners differ in a number of ways which affect L2 acquisition, in particular
their rate of development and their ultimate level of achievement. The ways in
which learners differ are potentially infinite as they reflect the whole range of
variables relating to the cognitive, affective, and social aspects of a human
being. The factors can be divided into those that are fixed and immutable and
those that are variable, influenced by social setting and by the actual course of
L2 development. This distinction is, perhaps, best viewed as a continuum
with specific factors treated as more or less immutable/mutable.
Age is an example of a fixed factor, in the sense that it is beyond external
control. Like many other issues, the role of age in L2 acquisition is controver¬
sial. The controversy centres around whether there is a critical period for L2
acquisition and, if so, when it ends. Long (1990a) has presented evidence to
suggest that the acquisition of a native-like accent is not possible by learners
36 Background

who begin learning after 6 years of age. He also argues that it is very difficult
for learners who begin at puberty to acquire native-like grammatical compet¬
ence. However, Scovel (1988) has presented somewhat different evidence to
argue that the critical period for a native-like pronunciation is around 12
years old. He claims that the evidence in favour of a critical period for gram¬
mar is equivocal (‘a potential maybe to a probable no’). There is general
agreement, however, that older learners enjoy an initial advantage in rate of
acquisition. A key theoretical issue relating to the age issue and the reason
why it has attracted considerable attention is whether adult L2 learners have
continued access to the innate knowledge of linguistic universals (see above)
which guide children’s acquisition of their mother tongue.
It has also been claimed that language learning aptitude constitutes a relat¬
ively immutable factor (Carroll 1981). This refers to the specific ability for
language learning which learners are hypothesized to possess. Much of the
early work on aptitude focused on developing tests to measure it. Carroll and
Sapon (1959) developed the Modern Language Aptitude Test (MLAT) and
Pimsleur (1966) developed the Pimsleur Language Aptitude Battery. These
tests conceptualized aptitude in modular form. Different modules measured
such skills as the learner’s ability to perceive and memorize new sounds, to
identify syntactic patterns in a new language, to detect similarities and differ¬
ences in form and meaning, and to relate sounds to written symbols. The em¬
pirical work involved in developing these tests was very rigorous and it is a
pity that it has not been followed up more extensively in recent years. A recent
study by Skehan (1990) indicates the importance of aptitude as an explanat¬
ory factor for both LI and L2 learning.
Motivation is an example of a factor that is clearly variable. The strength of
an individual learner’s motivation can change over time and is influenced by
external factors. There is widespread recognition that motivation is of great
importance for successful L2 acquisition, but there is less agreement about
what motivation actually consists of. Motivation can be causative (i.e. have
an effect on learning) and it can be resultative (i.e. be influenced by learning).
It can be intrinsic (i.e. derive from the personal interests and inner needs of the
learner) and it can be extrinsic (i.e. derive from external sources such as ma¬
terial rewards). There is a rich literature on motivation in general psychology,
which has not been fully exploited in SLA. The main body of work in SLA re¬
search is that associated with Gardner, Lambert, and their associates (see
Gardner 1985 for a summary). This is based on the assumption that the main
determinants of motivation are the learners’ attitudes to the target language
community and their need to learn the L2. Motivation, so measured, affects
the extent to which individual learners persevere in learning the L2, the kinds
of learning behaviours they employ (for example, their level of participation
in the classroom), and their actual achievement. Recent discussions of mo¬
tivation (for example, Crookes and Schmidt 1990), however, have emphas-
 Second language acquisition research: An overview 37

ized the need for investigating other aspects of motivation in L2 learning such
as intrinsic motivation.
Other individual learner factors lie somewhere between the two poles of
the continuum. One such factor that has attracted considerable attention, al¬
though with rather uncertain outcomes, is cognitive style. This is the term
used to refer to the way people perceive, conceptualize, organize, and recall
information. Various dimensions of cognitive style have been identified in
psychology but the one that has attracted the most attention in SLA is field
dependence/independence. Field-dependent learners operate holistically,
whereas field-independent learners are analytic. The main research hypo¬
thesis is that field-independent learners will be more successful at formal,
classroom learning, but the studies completed to date provide only limited
support for this hypothesis (see Griffiths and Sheen 1992).
Chapter 11 will consider individual learner differences, concentrating on
the above factors, but also dealing briefly with a number of others.

Learner strategies

Learner strategies are conscious or potentially conscious; they represent the

learner’s deliberate attempts to learn. Oxford (1989) defines them as ‘beha¬
viors or actions which learners use to make language learning more success¬
ful, self-directed and enjoyable’. The vagueness of this and other definitions
points to a major problem in this area of SLA research—how to identify, de¬
scribe, and classify the ‘behaviors and actions’ that constitute learners’ at¬
tempts to learn. Nevertheless, the study of learner strategies has been one of
the main areas of growth in SLA research (see O’Malley and Chamot 1989
and Oxford 1990).
A distinction is often drawn between learner strategies that are cognitive
(for example, relating new concepts to other information in memory), those
that are metacognitive (for example, organizing a personal timetable to facil¬
itate effective study of the L2) and those that are social (for example, seeking
out opportunities to converse with native speakers).
There have been a number of attempts to establish which learner strategies
facilitate L2 acquisition. Studies of the good language learner (for example,
Naiman, Frohlich, Stern, and Todesco 1978) have tried to identify the strate¬
gies which successful learners use. Experimental studies (for example,
Politzer and McGroarty 1985) have tried to establish whether the use of
‘good learning behaviors’ does lead to gains in learning. There have also been
attempts to train learners to use effective strategies (for example, O’Malley
1987).
Learner strategies are examined in Chapter 12, while attempts to train
strategy use are considered in Chapter 14.
38 Background

Summary and conclusion

This chapter began by defining the term Second Language Acquisition, point¬
ing out the diversity of phenomena which have been investigated under its
banner. It then went on to identify four general questions which SLA research
has addressed. A framework for reviewing SLA research was then proposed.
This consisted of four major areas: (1) learner language, (2) learner-external
factors, (3) learner-internal factors, and (4) the language learner as an indi¬
vidual. A brief survey of some of the main issues relating to each of these areas
was then provided.
It is useful to distinguish two branches of enquiry within SLA research. One
has as its focus learning and the other the language learner. In the case of the
former the emphasis is on identifying the universal characteristics of L2 ac¬
quisition. In the case of the latter the aim is to account for differences in ways
in which individual learners learn an L2. These two branches have tended to
work independently of each other. A complete theory of SLA must account
for and interrelate the findings from both branches.
It is also useful to distinguish two main goals of SLA research: description
and explanation. In the case of description, the goal is to provide a clear and
accurate account of the learner’s competence and, in particular, to uncover
the regularities and systematicities in the learner’s development and control
of L2 knowledge. In the case of explanation, one goal is to reveal how learners
are able to develop knowledge of an L2 from the available input and how they
use this knowledge in communication. A second goal is to specify the factors
that cause variation in individual learners’ accomplishment of this task. A
complete theory of SLA must provide an explanation for what is currently
known about learner language and about language learners.

Notes
1 The claim that there are some properties of a language that are entirely for¬
mal in nature is not uncontroversial. However, it is an assumption of lin¬
guists working within the generative tradition of Chomsky that the nature
of many syntactical rules is determined by highly abstract principles of a
purely formal kind (see Chapter 10).
2 A less problematic distinction than ‘acquisition/learning’ is that between
‘implicit’ and ‘explicit’ knowledge. This is because this distinction does not
rest on how the knowledge was internalized but on how L2 knowledge is
represented and used by learners. The implicit/explicit distinction is exam¬
ined in Chapter 9.
3 The focus on naturalistic learners came about because early research exam¬
ined ‘second’ rather than ‘foreign’ language learners—L2 learners of Eng¬
lish in the United States or the United Kingdom and L2 learners of German
in Germany, for example. It should be noted, however, that many of these
 Second language acquisition research: An overview 39

‘natural’ learners were also receiving some instruction. A better label,

therefore, might be ‘mixed’.
4 A distinction is generally drawn between those cognitive processes res¬
ponsible for the general pattern of L2 acquisition (i.e. its universal charac¬
teristics) and those strategies that are employed, often consciously, by
learners to improve their learning. The terminological distinction between
‘processes’ and ‘strategies’ is not always adhered to, however. ‘Strategies’
are often invoked to account for developmental regularities. The problems
of definition are dealt with in Chapter 12.
5 The claims made on behalf of Parallel Distributed Processing are not in¬
compatible with a number of other theoretical positions. The notion of
‘connection strengths’, for example, is compatible with the idea of
‘weights’ found in the Competition Model and also relates closely to work
done in variability in learner language.
6 It has been suggested that in the case of principles with multiple para¬
meters, learners first choose an unmarked parameter which can be sub¬
sequently abandoned solely on the basis of the positive evidence available
to them in the input. They do not initially choose a marked parameter
which can only be abandoned with the help of negative evidence (for ex¬
ample, formal corrections), as this is generally not available to them in the
input. They resort to a marked setting only when unmarked settings have
been tried and dismissed (see White 1989a, Chapter 6, for a detailed
discussion).

Further reading
Increasingly, work in SLA research deals with highly specific issues, or altern¬
atively presents a particular view of the process of acquisition. There are few
overviews written from an objective standpoint and the task of writing such
overviews becomes more difficult as the field grows more complex.
R. Ellis, Understanding Second Language Acquisition (Oxford University
Press, 1985) provides an account of much of the earlier research and theoriz¬
ing.
L. Beebe (ed.), Issues in Second Language Acquisition: Multiple Perspectives
(Newbury House, 1988), contains papers surveying a number of the key
areas, including one not dealt with in this book—the neurolinguistic
perspective.
D. Larsen-Freeman and M. Long, An Introduction to Second Language Ac¬
quisition Research (Longman, 1991), provides a more comprehensive and
up-to-date account of most of the areas mentioned in this chapter. It also pro¬
vides an extremely useful account of methodology in SLA research.
The Annual Review of Applied Linguistics, Volume 9, contains a number of
articles surveying many of the areas of SLA research referred to in this
chapter.
40 Background

For readers who do not have the time to consult full-length books two art¬
icles provide short and accessible accounts of some of the areas:
C. James, ‘Learner language’, Language Teaching Abstracts (1990) 23:
205-213.
D. Larsen-Freeman (1991), ‘Second language acquisition research: Staking
out the territory’, TESOL Quarterly (1992) 25: 315-50.
One way of starting the study of SLA, however, is by reading some of the
early case studies:
E. Hatch (ed.), Second Language Acquisition, (Newbury House, 1978a),
includes a number of such studies (see in particular the papers by Huang and
Hatch, Cancino et ah, and Butterworth and Hatch).
Schmidt’s study of one adult learner called Wes is perhaps the best known
and most rewarding of the case studies. See:
R. Schmidt, ‘Interaction, acculturation, and the acquisition of communicat¬
ive competence: A case study’ in N. Wolfson and E. Judd (eds.), Socio¬
linguistics and Language Acquisition (Newbury House, 1983).
PART TWO
The description of learner language
 '

t
Introduction

An obvious starting point in the study of second language (L2) acquisition is

the study of the language that learners produce at different stages of their de¬
velopment. Learner language can provide the researcher with insights into the
process of acquisition. For many researchers—although not all1—it consti¬
tutes the most important source of information about how learners learn an
L2. This section, therefore, focuses on the description of learner language. Its
purpose is to provide an overview of the principal ways in which it has been
studied and to describe its main characteristics.
A number of different approaches to the description of learner language
can be identified:

1 the study of learners’ errors

2 the study of developmental patterns
3 the study of variability
4 the study of pragmatic features.

In the relatively short history of SLA research, it is possible to see a progres¬

sion in the way enquiry has proceeded. Thus, initially the main approach was
the study of learners’ errors, but this was rapidly superseded by the study of
developmental patterns and, a little later, variability. The study of L2 prag¬
matic features is a more recent phenomenon. This section follows this histor¬
ical progression, beginning with (1) and concluding with (4). The opening
chapters, therefore, focus on the early research, while the later chapters give
more attention to more current research. However, as all these approaches
have continued to figure in SLA research at least to some extent, each chapter,
including the early ones, will also examine relevant recent research.
As we noted in the Introduction to this book, prior to the late 1960s there
was almost no empirical study of L2 acquisition. Why, then, did researchers
suddenly become interested in it? There appear to have been two principal
reasons. One concerned the need to investigate the claims of competing theo¬
ries. The other concerned the desire to improve L2 pedagogy.
According to the Contrastive Analysis Hypothesis, learners were strongly
influenced by their LI. Where the LI matched the L2, learning was facilitated;
where it differed, learning was impeded. In the view of some (for example,
Lado 1957), errors were mainly, if not entirely, the result of transfer of LI
‘habits’. This theory of learning was challenged both by Chomsky’s attack on
behaviourism (see his famous review of B. Skinner’s Verbal Learning 'm 1958)
44 The description of learner language

and also by research on LI acquisition, which showed that children did not
seem to learn their mother tongue as a set of ‘habits’ but rather seemed to con¬
struct mental ‘rules’, which often bore no resemblance to those manifest in
their caretakers’ speech. This challenge to the received opinion of the day cre¬
ated the necessary climate for the empirical study of L2 acquisition. Were
learners’ errors the result of LI transfer? Did L2 learners, like LI learners,
construct unique mental ‘rules’? These were questions that could only be an¬
swered by looking at learner language.
Second, many L2 researchers were directly concerned with language ped¬
agogy. The prevailing methods of the day were the audiolingual method and
the oral/situational approach (see Richards and Rogers 1986). Both of these
emphasized tightly structuring the input to the learner and controlling output
in order to minimize errors. It was noted, however, that children were suc¬
cessful in acquiring their mother tongue without such a structured learning
environment. Also, many L2 learners, children and adults, seemed to be very
successful in learning an L2 in natural settings. Newmark (1966), in a seminal
paper, argued that L2 learning in the classroom would proceed more effici¬
ently if teachers stopped ‘interfering’ in the learning process. But how did L2
learners learn in natural settings? What strategies did they use? What made
some learners more successful than others? Again, these were questions that
invited empirical enquiry. Many of the early studies in SLA research investig¬
ated L2 learners in untutored settings, motivated in part by the desire to find
what experiences worked for them, so that suitable copies could be intro¬
duced into the language classroom.
This, then, constitutes the background to the study of learner language. It
should be clear that although the empirical studies were primarily descriptive
in nature, they were not atheoretical. In the late 1960s and 1970s a growing
consensus was reached that behaviourist theories of L2 learning were inad¬
equate. L2 learners, like LI learners, were credited with a ‘built-in-syllabus’
(Corder 1967), which guided their progress. Selinker (1969,1972) coined the
term ‘interlanguage’ to refer to the special mental grammars that learners
constructed during the course of their development. Interlanguage theory
credited learners with playing an active role in constructing these grammars.
It treated their behaviour, including their errors, as rule-governed. The lan¬
guage they produced, therefore, reflected the strategies they used to construct
provisional grammatical rules (i.e. rules which they subsequently revised).
The research that we will now consider helped to shape interlanguage theory
and, in many cases, was influenced by it. Thus, although the main concern of
this section is the description of learner language, some attention to theory is
inevitable. Interlanguage theory is considered more fully in Chapter 9.
The research reported in these chapters was often conducted by researchers
whose primary concern was L2 acquisition and L2 pedagogy. Inevitably,
however, these researchers drew on various fields to help them in their en¬
quiry. The fields of most obvious importance were linguistics (which
 The description of learner language 45

provided researchers with well-defined linguistic categories to investigate),

first language acquisition research (which provided them with useful proced¬
ures for collecting and analysing learner language), sociolinguistics (which
provided both theories and methodologies for examining variability), and
pragmatics/discourse analysis (which facilitated the study of learner language
in a wider social and textual context). The account of learner language that
follows, therefore, reflects one of the prevailing characteristics of SLA re¬
search—the utilization of a wide range of concepts and methods that have
been borrowed from different fields and adapted to the particular needs of L2
researchers.

Notes
1 Some researchers prefer to work with intuitional data, obtained by asking
learners to judge the grammaticality of sentences. However, as this intro¬
duction to Section 2 makes clear, the primary data in SLA research—both
historically speaking and in terms of importance—has been learner
language.
V

f
2 Learner errors and error analysis

Introduction

Learners make errors in both comprehension and production. An example of

a comprehension error is when a learner misunderstands the sentence1 ‘Pass
me the paper’ as ‘Pass me the pepper’, because of an inability to discriminate
the sounds /ei/ and Id. However, comprehension errors have received scant
attention, for, as Corder (1974: 125) has pointed out, although we can test
comprehension in general terms, ‘it is very difficult to assign the cause of fail¬
ures of comprehension to an inadequate knowledge of a particular syntactic
feature of a misunderstood utterance’. There is, in fact, a fundamental differ¬
ence between comprehension and production in processing terms, a point
that will be taken up later in Chapter 7. In this chapter, in accordance with the
main focus in second language acquisition research, we will concentrate on
production errors.
L2 learners are not alone in making errors. Children learning their first lan¬
guage (LI) also make ‘errors’ (for example, Bloom 1970). They regularly
produce utterances like the following:
*1 goes see Auntie May. (= I went to see Auntie May.)
"'Eating ice cream. (= I want to eat an ice cream.)
"'No writing in book. (= Don’t write in the book.)
Also, even adult native speakers sometimes make ‘errors’.2 For example, they
may sometimes omit a grammatical morpheme as in:
"'My father live in Gloucester. (= My father lives in Gloucester.)
But it is probably true to say that these ‘errors’ are not generally thought of as
errors in the same sense as those produced by L2 learners. Whereas L2
learners’ errors are generally viewed as ‘unwanted forms’ (George 1972),
children’s ‘errors’ are seen as ‘transitional forms’ and adult native speakers’
errors as ‘slips of the tongue’.
The study of errors is carried out by means of Error Analysis (EA). In the
1970s, EA supplanted Contrastive Analysis (CA), which sought to predict the
errors that learners make by identifying the linguistic differences between
their LI and the target language.3 The underlying assumption of CA was that
errors occurred primarily as a result of interference when the learner trans¬
ferred native language ‘habits’ into the L2. Interference was believed to take
place whenever the ‘habits’ of the native languages differed from those of the
48 The description of learner language

target language. CA gave way to EA as this assumption came to be chal¬

lenged. Whereas CA looked at only the learner’s native language and the tar¬
get language (i.e. fully-formed languages), EA provided a methodology for
investigating learner language. For this reason EA constitutes an appropriate
starting point for the study of learner language and L2 acquisition.
EA was, of course, not a new development. The analysis of learner errors
had long been a part of language pedagogy. French (1949), for example, pro¬
vides a comprehensive account of common learner errors. Lee (1957) reports
on an analysis of some 2,000 errors in the written work of Czechoslovakian
learners, which were ‘hurriedly grouped into categories’ (wrong punctuation,
misuse, or omission of articles, misspellings, non-English constructions, and
wrong use of tenses). Lee argued that such an analysis put the teacher in ‘a
better position to decide how teaching time should be spent.’ But such tradi¬
tional analyses lacked both a rigorous methodology and a theoretical frame¬
work for explaining the role played by errors in the process of L2 acquisition.
It was not until the 1970s that EA became a recognised part of applied lin¬
guistics, a development that owed much to the work of Corder (see Corder
1981a for a collection of his papers).
In an early, seminal article, Corder (1967) noted that errors could be signi¬
ficant in three ways: (1) they provided the teacher with information about
how much the learner had learnt, (2) they provided the researcher with evid¬
ence of how language was learnt, and (3) they served as devices by which the
learner discovered the rules of the target language. Whereas (1) reflects the
traditional role of EA, (2) provides a new role that is of primary interest to the
L2 researcher because it could shed light on (3)—the process of L2 acquisi¬
tion. It should be noted, though, that many of the researchers who carried out
error analyses in the 1970s continued to be concerned with language teach¬
ing. Indeed, the attempt to discover more about L2 acquisition through the
study of errors was itself motivated by a desire to improve pedagogy.
Corder (1974) suggests the following steps in EA research:

1 Collection of a sample of learner language

2 Identification of errors
3 Description of errors
4 Explanation of errors
5 Evaluation of errors.

However, many studies do not include Step 5 and, in fact, the evaluation of
learner errors has generally been handled as a separate issue, with its own
methods of enquiry. Error evaluation studies are considered in this chapter as
they represent one of the ways in which EA developed in the 1970s and
1980s.
This chapter will consider the procedures involved in each of these steps. In
so doing, it will examine some of the research carried out in the 1970s and,
where appropriate, the methodological problems. A general critique of EA
 Learner errors and error analysis 49

follows. Finally, it will look at some more recent research which has made use
of the techniques of error analysis.

Collection of a sample of learner language

The starting point in EA is deciding what samples of learner language to use

for the analysis and how to collect these samples.
We can identify three broad types of EA according to the size of the sample.
A massive sample involves collecting several samples of language use from a
large number of learners in order to compile a comprehensive list of errors,
representative of the entire population. A specific sample consists of one sam¬
ple of language use collected from a limited number of learners, while an in¬
cidental sample involves only one sample of language use produced by a
single learner. Clearly an EA based on a massive sample is a major
undertaking and it is not surprising that most published EAs have employed
specific or incidental samples.4
The errors that learners make can be influenced by a variety of factors. For
example, they may make errors in speaking, but not in writing, as a result of
the different processing conditions involved. Learners with one LI may make
a particular error which learners with a different LI do not. This points to the
importance of collecting well-defined samples of learner language so that
clear statements can be made regarding what kinds of errors the learners pro¬
duce and under what conditions. Table 2.1 lists some of the factors that need
to be considered to ensure this. Unfortunately, many EA studies have not paid
sufficient attention to these factors, with the result that they are difficult to in¬
terpret and almost impossible to replicate. Table 2.1 is not complete; the fac¬
tors that can bring about variation in learner output are numerous, perhaps

Factors Description

A Language

Medium Learner production can be oral or written

Genre Learner production may take the form of a conversation,
a lecture, an essay, a letter, etc.
Content The topic the learner is communicating about

B Learner

Level Elementary, intermediate, or advanced

Mother tongue The learner’s LI
Language learning experience This may be classroom or naturalistic or a mixture of the
two

Table 2.1: Factors to consider when collecting samples of learner language

50 The description of learner language

infinite. Little attention was paid to them in the early 1970s (see Chapter 4),
though they were taken up in later work.
Decisions also need to be made regarding the manner in which the samples
are to be collected. An important distinction is whether the learner language
reflects natural, spontaneous language use, or is elicited in some way. Natural
samples are generally preferred. A drawback, however, is that learners often
do not produce much spontaneous data, which led Corder (1973) to argue
the case for elicited data. Elicitation, however, is not to be confused with test¬
ing, which is concerned with measuring the learner’s knowledge for purposes
of evaluating rather than describing competence. Corder identifies two kinds
of elicitation. Clinical elicitation involves getting the informant to produce
data of any sort, for example, by means of a general interview or by asking
learners to write a composition. Experimental methods involve the use of spe¬
cial instruments designed to elicit data containing the linguistic features
which the researcher wishes to investigate. An example of such an elicitation
instrument is the Bilingual Syntax Measure (Burt, Dulay, and Hernandez
1973). This consisted of a series of pictures which had been devised to elicit
specific features and which the learners were asked to describe. The authors
claimed that the resulting corpus of language reflected natural speech. The
method of data collection can have a marked effect on the results obtained, as
a result of the different production processes which they typically involve.
Lococo (1976) found differences in the number and type of errors in samples
of learner language collected by means of free composition, translation, and
picture composition. For example, errors reflecting LI influence were, not
surprisingly, more common in the translation task. Again, the effects of task
on learner language are considered in detail in Chapter 4.
Another issue is whether the samples of learner language are collected
cross-sectionally (i.e. at a single point in time) or longitudinally (i.e. at suc¬
cessive points over a period of time). The majority of EAs have been cross-
sectional, thus making it difficult to determine accurately the different errors
that learners produce at different stages of their development.
The limitations of EA, as practised in the late 1960s and 1970s, are evident
in the samples of learner language collected. Svartvik, for instance, notes that
‘most error analyses use regular examination papers (composition, transla¬
tions, etc.) for material’ (1973b: 12). There were few studies of learner
speech. Also, as we have noted, insufficient attention was paid to identifying
and controlling the factors that might potentially influence the errors that
learners produced.

Identification of errors

Once a corpus of learner language has been collected, the errors in the corpus
have to be identified. It is necessary to decide, therefore, what constitutes an
‘error’ and to establish a procedure for recognizing one.
 Learner errors and error analysis 51

An error can be defined as a deviation from the norms of the target lan¬
guage. This definition raises a number of questions, however. First, there is
the question regarding which variety of the target language should serve as
the norm. The general practice, especially where classroom learners are con¬
cerned, is to select the standard written dialect as the norm. This, of course, is
fundamentally wrong if the goal is to describe learners’ oral production. Nor
is it always possible to adopt the standard spoken variety as the norm. Some
learners are exposed to varieties of the language which differ from the stand¬
ard dialect. For example, in comparison with the norms of British or Amer¬
ican standard written English the utterance

'"She coped up with her problem very well.

is erroneous, but in comparison with norms of educated Zambian English

such an utterance can be considered correct/
A second question concerns the distinction between errors and mistakes
(Corder 1967). An error (in this technical sense) takes place when the devi¬
ation arises as a result of lack of knowledge. It represents a lack of compet¬
ence (see Chapter 1, page 12). A mistake occurs when learners fail to perform
their competence. That is, it is the result of processing problems that prevent
learners from accessing their knowledge of a target language rule and cause
them to fall back on some alternative, non-standard rule that they find easier
to access. Mistakes, then, are performance phenomena and are, of course,
regular features of native-speaker speech, reflecting processing failures that
arise as a result of competing plans, memory limitations, and lack of auto-
maticity. Corder argues that the EA should be restricted to the study of errors
(i.e. mistakes should be eliminated from the analysis). However, apart from
the problems of identification that this distinction raises, it also assumes that
competence is homogeneous rather than variable. Thus, if learners sometimes
use a correct target form and sometimes an incorrect, non-target form, it can¬
not necessarily be concluded that the learner ‘knows’ the target form and that
the use of the non-target form represents a mistake. It is possible that the
learner’s knowledge of the target form is only partial; the learner may not
have learnt all the contexts in which the form in question can be used. For
example, a learner may have no difficulty in using the target language form in
some linguistic contexts:

My sisters are older than me.

but produce an error in others:

*My three sister are older than me.

In this early period, the study of learner errors largely ignored the problem of
variability in learner language (see Chapter 4).
52 The description of learner language

A third question concerns whether the error is overt or covert (Corder

1971a). An overt error is easy to identify because there is a clear deviation in
form, as when a learner says:

*1 runned all the way.

A covert error occurs in utterances that are superficially well-formed but

which do not mean what the learner intended them to mean. For example, the
utterance (from Corder 1971a):

*It was stopped.

is apparently grammatical until it becomes clear that ‘it’ refers to ‘the wind’.
Furthermore, a superficially correct utterance may only be correct by chance.
For example, the learner may manifest target-like control of negative con¬
structions in ready-made chunks such as ‘I don’t know’ but fail to do so in
‘created’ utterances (i.e. utterances that are constructed on the basis of rules
the learner has internalized). The existence of covert errors led Corder to ar¬
gue that ‘every sentence is to be regarded as idiosyncratic until shown to be
otherwise’ (page 21).
A fourth question concerns whether the analysis should examine only devi¬
ations in correctness or also deviations in appropriateness. The former in¬
volves rules of usage and is illustrated in the two examples above. The latter
involves rules of language use. For example, a learner who invites a relative
stranger by saying ‘I want you to come to the cinema with me’ has succeeded
in using the code correctly but has failed to use it appropriately. In general,
EA has attended to ‘breaches of the code’ and ignored ‘misuse of the code’
(Corder 1974: 124), but more recently attention has been paid to the latter
(Thomas 1983). Errors of this pragmatic kind will be considered in
Chapter 5.
These various distinctions are indicative of the kinds of problems which
analysts face in recognizing errors. To overcome them, Corder (1971a; 1974)
proposes an elaborate procedure for identifying errors, which is shown in Fig¬
ure 2.1. This procedure acknowledges the importance of ‘interpretation’ and
distinguishes three types: normal, authoritative, and plausible. A normal in¬
terpretation occurs when the analyst is able to assign a meaning to an utter¬
ance on the basis of the rules of the target language. In such cases, the
utterance is ‘not apparently erroneous’, although it may still only be right ‘by
chance’. An authoritative interpretation involves asking the learner (if avail¬
able) to say what the utterance means and, by so doing, to make an ‘authorit¬
ative reconstruction’. A plausible interpretation can be obtained by referring
to the context in which the utterance was produced or by translating the sen¬
tence literally into the learner’s LI.
There are a number of major methodological problems with the proced¬
ures used in error identification, some of which we have already noted.
Corder’s (1967) distinction between errors and mistakes is not easy to put
 Learner errors and error analysis
53

Figure 2.1: Algorithm for providing for description of idiosyncratic dialects (from Corder 1981a)
54 The description of learner language

into practice, even if the learner is available to provide an ‘authoritative inter¬

pretation’. In addition, the distinction does not take account of the possibility
that learners’ knowledge is variable. Also, it is not at all clear whether
Corder’s suggestions for identifying covert errors will work. Reliance on the
learner as an informant has been criticized on the grounds that retrospective
accounts of intended meaning are often not reliable (Van Els et al. 1984). Also,
such a procedure assumes that learners possess the necessary metalingual
knowledge to talk about their own performance—an assumption that may
not be justified in the case of children and some adult learners. Not all re¬
searchers have found problems with identifying errors, however. Duskova
(1969), in a study considered in the next section, discusses it at some length
and concludes that ‘the number of cases in which it was hard to decide
whether an error had been made ... did not exceed 4 per cent of all the errors
examined’. Unfortunately, Duskova does not provide inter-rater reliability
measures for the errors identified in her sample—a failing common to most of
the early studies.6

Description of errors
The description of learner errors involves a comparison of the learner’s idio¬
syncratic utterances with a reconstruction of those utterances in the target
language. It requires, therefore, attention to the surface properties of the
learners’ utterances (i.e. it does not attempt, at this stage, to identify the
sources of the errors). Some researchers have felt the need to maintain a clear
distinction between the description and explanation of errors. Dulay, Burt,
and Krashen (1982), for example, argue the need for descriptive taxonomies
of errors that focus only on observable, surface features of errors, as a basis
for subsequent explanation.
Perhaps the simplest type of descriptive taxonomy is one based on lin¬
guistic categories. This type is closely associated with a traditional EA under¬
taken for pedagogic purposes, as the linguistic categories can be chosen to
correspond closely to those found in structural syllabuses and language text
books. An example can be found in Burt and Kiparsky’s The Gooficon: A Re¬
pair Manual for English (1972). This identifies a number of general linguistic
categories (for example, the skeleton of English clauses, the auxiliary system,
passive sentences, temporal conjunctions, and sentential complements). Each
general category is then broken down into further levels of subcategories. For
example, the auxiliary system is subdivided into ‘do’, ‘have and be’, modals,
and mismatching auxiliaries in tag questions, while errors in the use of ‘do’
are classified according to whether they involve over-use in questions and
negatives, underuse in questions or overuse in affirmative sentences. Politzer
and Ramirez (1973) begin with even more general categories: morphology,
syntax, and vocabulary. Such taxonomies allow for both a detailed descrip¬
tion of specific errors and also for a quantification of a corpus of errors.
 Learner errors and error analysis 55

The 1960s saw a number of studies which provided descriptions of the dif¬
ferent kinds of linguistic errors produced by learners. Richards (1971b), in a
paper designed to challenge the widely held belief that learner errors were the
result of LI interference, provided a taxonomy of different categories of lin¬
guistic error based on a number of previous studies. He examined errors
made by learners from different language backgrounds (Japanese, Chinese,
Burmese, French, Czech, Polish, Tagalog, Maori, Maltese, and the major In¬
dian and West African languages) and illustrated the different kinds of errors
relating to the production and distribution of verb groups, prepositions, art¬
icles, and the use of questions. However, he made no attempt to quantify the
errors. Nor do we know to what extent his linguistic categories accounted for
all the errors he examined. Duskova (1969)—one of the studies Richards
drew on—is better in this respect. She identified a total of 1007 errors in the
written work of 50 Czech learners of English, who were postgraduate stu¬
dents studying science. She found 756 ‘recurrent systemic errors’ and 251
‘nonce errors’ (i.e. errors that occurred once only). Errors in articles were
most common (260), followed by errors in lexis (233) and morphology (180).
In comparison, there were only 54 errors in syntax and 31 in word order.
Duskova noted, however, that the frequency of the errors did not necessarily
reflect the level of difficulty the learners experienced with different linguistic
features, as some features (such as articles) were attempted more often than
others (for example, adverbs). Duskova also noted that although she had few
difficulties in assigning errors to general linguistic categories such as ‘word
order’, it often proved very difficult to classify them accurately into
subcategories.
These studies were cross-sectional in design. Of greater interest for SLA re¬
search are longitudinal studies of learners’ errors as these can show in what
areas of language errors persist over time. Chamot’s (1978; 1979) study of
the acquisition of English by a bilingual French/Spanish boy is interesting in
this respect. She found that the main linguistic problem areas were omission
of constituents, verb forms, sentence formation, articles, and prepositions. In
some of these (for example, omission of constituents) the number of errors re¬
duced sharply over a 44-month period, while in others (for example, question
formation) little improvement was evident. In all the areas, however, there
was considerable fluctuation in error frequency throughout the period.
Chamot’s study suggests that it may be difficult to provide a satisfactory de¬
scription of learners’ L2 development by quantifying the types of errors they
make.
An alternative to a linguistic classification of errors is to use a surface stra¬
tegy taxonomy. This ‘highlights the ways surface structures are altered’
(Dulay, Burt, and Krashen 1982: 150) by means of such operations as omis¬
sions, additions and regularizations. Table 2.2 provides a part of the total
taxonomy together with examples of the categories. Dulay, Burt, and
Krashen claim that such an approach is promising because it provides an
56 The description of learner language

indication of the cognitive processes that underlie the learner’s reconstruction

of the L2. This seems a doubtful claim, however, as it presupposes that learn¬
ers operate on the surface structures of the target language rather than create
their own, unique structures. If a surface strategy taxonomy does not repres¬
ent mental processes, it is not clear what value it has. This may account for
why there have been few attempts to describe learner errors using such a
taxonomy.

Category Description Example

Omissions The absence of an item that must She sleeping.

appear in a well-fbrmed utterance.
Additions The presence of an item that must not We didn’t went there.
appear in well-formed utterances.
Misinformations The use of the wrong form of the The dog ated the chicken.
morpheme or structure.
Misorderings The incorrect placement of a What daddy is doing?
morpheme or group of morphemes in
an utterance.

Table 2.2: A surface strategy taxonomy of errors (categories and examples taken
from Dulay, Burt, and Krasben 1982)

Although linguistic and surface strategy taxonomies of errors may have a

pedagogic application (for example, by demonstrating which errors are the
most frequent and, therefore, most in need of attention), in general they shed
little light on how learners learn an L2. Corder’s (1974) framework for
describing errors is more promising in this respect. He distinguishes three
types of error according to their systematicity:

1 Presystematic errors occur when the learner is unaware of the existence of a

particular rule in the target language. These are random.
2 Systematic errors occur when the learner has discovered a rule but it is the
wrong one.
3 Postsystematic errors occur when the learner knows the correct target lan¬
guage rule but uses it inconsistently (i.e. makes a mistake).

In order to identify these different kinds of errors, however, it is necessary to

interview the learner. Thus, type (1) occurs when the learner cannot give any
account of why a particular form is chosen, type (2) occurs when the learner is
unable to correct the errors but can explain the mistaken rule used, and type
(3) occurs when the learner can explain the target-language rule that is norm¬
ally used. Such a taxonomy, therefore, requires that the researcher has access
to the learners and that the learners are capable of providing explanations for
their L2 behaviour. For this reason it may prove difficult to operate.
 Learner errors and error analysis 57

The description of errors, like their identification, is problematic. Even if

the error itself can be easily identified, it is often problematic to determine
what the error consists of. If a learner produces the following sentence:

*My name Alberto.

there is no difficulty in reconstructing the target-language version:

My name is Alberto.

and so establishing that copula ‘is’ has been omitted. But in many
cases—even with sentences that are overtly idiosyncratic—the reconstruc¬
tion of the target language version—and, therefore, its description—is prob¬
lematic. For example, if a learner produces the following sentence:

*1 am worried in my mind.

it is not clear what constitutes the best reconstruction. One possibility is ‘I am

feeling worried.’ Another is ‘I have a problem on my mind.’
Even if the learner is available for consultation, it may not be possible to
choose between these two reconstructions. But the description of the error
will obviously vary according to which reconstruction is finally chosen. The
reconstruction of covertly idiosyncratic sentences will prove even more diffi¬
cult.
Another problem concerns the failure to quantify the different types of
errors that have been identified and described. Many of the EA studies that
have been conducted have been very informal—perhaps as a result of the
kinds of problems discussed above. In some studies error frequencies are not
given at all (for example, Jain 1974; Richards 1971b), while in others only ab¬
solute frequencies are given (for example, Duskova 1969). But as Schachter
and Celce-Murcia (1977) point out, to say anything worthwhile about error
frequency we need to know the number of times it would be possible for
learners to have committed different errors. In other words, relative rather
than absolute frequencies are needed.

Explanation of errors
Assuming that it is possible to identify and describe errors, the next step is to
try to explain them. Explanation is concerned with establishing the source of
the error, i.e. accounting for why it was made. This stage is the most import¬
ant for SLA research as it involves an attempt to establish the processes res¬
ponsible for L2 acquisition.
As Taylor (1986) points out, the error source may be psycholinguistic,
sociolinguistic, epistemic, or may reside in the discourse structure. Psycho-
linguistic sources concern the nature of the L2 knowledge system and the
difficulties learners have in using it in production. Sociolinguistic
sources involve such matters as the learners’ ability to adjust their language in
58 The description of learner language

accordance with the social context. Epistemic sources concern the learners’
lack of world knowledge, while discourse sources involve problems in the or¬
ganization of information into a coherent ‘text’. In general, however, SLA re¬
search has attended only to the first of these. As Abbott puts it: ‘The aim of
any EA is to provide a psychological explanation’ (1980: 124). Figure 2.2
plots the different psycholinguistic sources to be discussed.

— transfer

competence intralingual (e.g. overgeneralization,

(‘errors’) transitional competence)

— unique (e.g. induced)

errors —

i— processing problems
performance _
(‘mistakes’)
— communication strategies

Figure 2.2: Psycholinguistic sources of errors

The distinction between ‘errors’ and ‘mistakes’, which has already been
discussed with regard to the identification stage of EA, is also relevant in ex¬
plaining deviations—demonstrating the interdependence of these two steps
in EA. Any deviation from target-language norms may reflect either a prob¬
lem in performance or in competence.7 It is helpful to recognize two different
kinds of performance mistake: those that result from processing problems of
various kinds, and those that result from such strategies as circumlocution
and paraphrase, which a learner uses to overcome lack of knowledge. The lat¬
ter are known as communication strategies and will be discussed in some de¬
tail in Chapter 9. As we have already seen, it is competence errors that have
been considered central to the study of L2 acquisition.
A number of different sources or causes of competence errors have been
identified. Richards (1971b) distinguishes three:

1 Interference errors occur as a result of ‘the use of elements from one lan¬
guage while speaking another.’ An example might be when a German
learner of L2 English says !:'‘I go not’ because the equivalent sentence in
German is ‘Ich gehe nicht’.
2 Intralingual errors ‘reflect the general characteristics of rule learning such
as faulty generalization, incomplete application of rules and failure to learn
conditions under which rules apply’.
3 Developmental errors occur when the learner attempts to build up hypoth¬
eses about the target language on the basis of limited experience.

However, Schachter and Celce-Murcia (1977) find the distinction between

intralingual and developmental errors ‘curious’, and most researchers have
 Learner errors and error analysis 59

operated with a general distinction between transfer errors (Richards’ cat¬

egory (1)) and intralingual errors (an amalgam of Richards’ (2) and (3)).
Transfer errors can be further subdivided. Lott (1983), for instance, distin¬
guishes three categories:

1 ‘Overextension of analogy’ occurs when the learner misuses an item

because it shares features with an item in the LI (for example, Italian learn¬
ers use ‘process’ to mean ‘trial’ because Italian ‘processo’ has this
meaning).
2 ‘Transfer of structure’ arises when the learner utilizes some LI feature
(phonological, lexical, grammatical, or pragmatic) rather than that of the
target language. This is what is generally understood as ‘transfer’.
3 ‘Interlingual/intralingual errors’ arise when a particular distinction does
not exist in the LI (for example, the use of ‘make’ instead of ‘do’ by Italian
learners because the ‘make/do’ distinction is non-existent in Italian).

It is this third category that has caused so many of the problems in de¬
termining whether an error is transfer or intralingual.
Intralingual errors are also often further subdivided. Thus, Richards
(1971b) distinguishes the following:

1 Overgeneralization errors arise when the learner creates a deviant struc¬

ture on the basis of other structures in the target language. It generally in¬
volves the creation of one deviant structure in place of two target language
structures (for example, “'‘He can sings’ where English allows ‘He can sing’
and ‘He sings’).
2 Ignorance of rule restrictions involves the application of rules to contexts
where they do not apply. An example is “ ‘He made me to rest’ through ex¬
tension of the pattern found with the majority of verbs that take infinitival
complements (for example, ‘He asked/wanted/invited me to go’).
3 Incomplete application of rules involves a failure to fully develop a struc¬
ture. Thus learners of L2 English have been observed to use declarative
word order in questions (for example, “'‘You like to sing?’) in place of inter¬
rogative word order (for example, ‘Do you like to sing?’). This type of
intralingual error corresponds to what is often referred to as an error of
transitional competence (Richards 1971a).
4 False concepts hypothesized (i.e. the learner fails to comprehend fully)
arise when the learner does not fully comprehend a distinction in the target
language—for example, the use of ‘was’ as a marker of past tense in “'‘One
day it was happened’.

It is not easy to distinguish transfer and intralingual errors, and even more
difficult to identify the different types of intralingual errors that Richards de¬
scribes. In an attempt to deal with the problem of identifying sources, Dulay
60 The description of learner language

and Burt (1974b) classified the errors they collected into three broad
categories:

1 Developmental (i.e. those errors that are similar to LI acquisition)

2 Interference (i.e. those errors that reflect the structure of the LI)
3 Unique (i.e. those errors that are neither developmental nor interference).

Dulay and Burt’s research has often been criticized on the grounds that reli¬
able classification of errors in terms of these categories is still not possible.
However, it might be argued that by using LI acquisition errors as a baseline
they at least provide an operational procedure for establishing which errors
are intralingual.
It is customary to distinguish another general source of errors. Induced
errors occur when learners are led to make errors by the nature of the instruc¬
tion they have received. Stenson (1974) provides a number of examples of
such errors in the classroom speech of Tunisian learners of English. Faulty ex¬
planation of grammatical points can give rise to errors (for example, the use
of ‘any’ to mean ‘none’ when the students were told that ‘any’ has a negative
meaning). Drills performed without consideration for meaning can also
result in error. Svartvik (1973b) suggests that overdrilling may be one of the
reasons why Swedish learners of L2 English overuse infinitival complements
(for example, " ‘He proposed her to stay’). Stenson argues that such errors are
not systematic and therefore do not reflect competence. However, there are
probably cases when learners do internalize faulty rules derived from instruc¬
tion and in such cases the resulting errors will reflect their competence. In¬
struction may constitute one source of what Dulay and Burt call ‘unique
errors’.
The bulk of the empirical work in SLA has focused on determining what
proportion of the total errors in a corpus are transfer as opposed to intra-
lingual. This is what motivated such studies as Richards (1971b) and Dulay
and Burt (1974b). The issue that these researchers tackled was the competing
claims of a behaviourist, habit-formation account of L2 acquisition and a
mentalist, creative-construction account. According to behaviourist ac¬
counts, errors were viewed as the result of the negative transfer of LI habits.
According to mentalist accounts, errors were predicted to be similar to those
found in LI acquisition because learners actively construct the grammar of an
L2 as they progress (i.e. they are intralingual). This issue dominated early
work in SLA. It should be noted, however, that subsequently researchers have
come to recognize that the correlation between behaviourism and transfer
errors on the one hand and mentalism and intralingual errors on the other is
simplistic and misleading. Transfer is now treated as a mental process in its
own right (see Chapter 8).
A good example of the kind of ‘proportion study’ that investigated this be-
haviourist/mentalist question is Dulay and Burt (1974b). Five hundred and
thirteen unambiguous errors produced by Spanish children acquiring L2
 Learner errors and error analysis 61

English were extracted from a corpus of speech collected by means of the

Bilingual Syntax Measure (see page 50 in this chapter). These errors occurred
in six syntactic structures which differed in English and Spanish. The errors
were classified as ‘developmental’, ‘interference’, or ‘unique’, and detailed
results for each structure were provided. In each case the developmental
errors far outweighed the interference errors. For example, for the structure
NP + V + Pronoun (for example, ‘The dog ate it’), which is realized as NP +
Pronoun + V in Spanish (for example, ‘El perro se lo comio’), there were 93
developmental errors and no interference errors. Overall less than 5 per cent
of the total errors were attributed to interference. This led Dulay and Burt to
propose that LI and L2 acquisition were very similar (the L2=L1 Hypo¬
thesis). It should be noted, however, that other studies (for example, George
1972 and Flick 1980) found a much higher proportion of transfer errors (see
Chapter 8).
One reason for the discrepancy in research findings has been the problem of
error classification. As Flick notes:

The assignment of a particular error to such categories as ‘transfer’, ‘over-

generalization’ or ‘ambiguous’ has been largely an arbitrary matter, sub¬
ject to the individual biases and point of view of the researcher (Flick
1979:60).
To solve this problem, Flick proposed the use of factor analysis. This is a stat¬
istical procedure which uses the patterns of correlation that exist among
scores on different variables to identify underlying factors. Flick reported on
the preliminary use of this procedure to identify clusters of errors in the
speech elicited from 20 adult Spanish learners of L2 English by means of an
oral translation task. Five factors emerged from the analysis:
1 transfer accounted for 34 per cent of the variance in the scores assigned to
each learner
2 performance (i.e. constraints imposed by processing load) for 23 per cent
3 simplifications of function words, such as errors of omission, for 17 per
cent
4 overgeneralization for 16 per cent
5 pronominal reference (for example, wrong choice of pronoun) for 11 per
cent.
Flick’s study represents an interesting attempt to increase the rigour of error
explanation, but it is open to a number of objections. For example, the five
categories identified constitute a very mixed bag: while some are explanatory
in nature, others seem more descriptive.
It is difficult to synthesize the results of attempts to explain errors in learner
language, but the following appear to be some of the main findings:
1 A large number—and in some cases perhaps most—of the errors that
learners produce are intralingual in origin rather than transfer.
62 The description of learner language

However, the precise proportion of the kinds of error varies considerably

from study to study.
2 According to Taylor (1975), learners at an elementary level produced
more transfer errors than learners at an intermediate or advanced level.
Conversely, he found that learners at an intermediate or advanced level
produced more intralingual errors (for example, overgeneralization) than
learners at an elementary level. However, as we will see in Chapter 8, other
researchers (for example, Kellerman 1983) have challenged the view that
transfer is more prevalent in beginners.
3 The proportion of transfer and intralingual errors varies in accordance
with the task used to elicit samples of learner language. Thus, translation
tasks tend to result in more transfer errors than tasks that call for free com¬
position (Lococo 1976).
4 Transfer errors are more common in the phonological and lexical levels of
language than in the grammatical level. Also some areas of grammar ac¬
quisition are more likely to be influenced by the learners’ LI than others.
Grauberg (1971) found that interference accounted for 25 per cent of the
lexical errors produced by adult German learners of L2 English, 10 per cent
of their syntactic errors, and none of their morphological errors.
5 Transfer errors are more common in adult learners than in child learners.
For example, White (1977) found that 21 per cent of the errors made by ad¬
ult Spanish learners of English were transfer. White used the same instru¬
ment to collect data as in the Dulay and Burt (1974b) study of Spanish
children referred to above, so this study is directly comparable.
6 Errors can have more than one source. For example, the ‘no’ + verb error
(as in *‘No look my card’) is universal, suggesting an intralingual explana¬
tion, but Spanish learners of L2 English have been noted to make this error
more frequently and for a longer period of time, suggesting that the LI pat¬
tern for negatives [no + verb) is also having an influence.

Again, it is important to recognize that the concept of ‘transfer’, upon which

many of the early EA studies were based, was simplistic. Transfer is, in fact, a
very complex notion which is best understood in terms of cognitive rather
than behaviourist models of learning.
It should be clear from this account of the explanation stage of EA that
problems abound. In particular, the concepts of transfer and intralingual
error were often not operationalized with sufficient rigour. Where one re¬
searcher identified the source of an error as transfer, another researcher
identified the source of the same error as intralingual. For example, Duskova
(1969) interpreted article deletion in Czech learners of English as interfer¬
ence, while Dulay and Burt (1974b) interpreted the same error in Spanish
children learning English as intralingual. Schachter and Celce-Murcia (1977)
argue that a large number of learners’ errors are ambiguous with regard to
source and that ‘one must be extremely cautious when claiming to have
 Learner errors and error analysis 63

identified the cause of any given error type’. Such caution has not always been
exercised, however. If clear explanatory statements about errors are often not
possible, the value of EA as a tool for investigating L2 acquisition is thrown
into question.

Evaluating errors

Whereas all the preceding stages of EA have involved an examination of

errors from the point of view of the learner who makes them, error evaluation
involves a consideration of the effect that errors have on the person(s) ad¬
dressed. This effect can be gauged either in terms of the addressee’s compre¬
hension of the learner’s meaning or in terms of the addressee’s affective
response to the errors. Error evaluation studies proliferated in the late 1970s
and in the 1980s, motivated quite explicitly by a desire to improve language
pedagogy.8 The studies surveyed in Table 2.3 constitute only a part of the
total. Ludwig (1982) provides a survey of twelve early studies.
The design of error evaluation studies involves decisions on who the ad¬
dressees (i.e. the judges) will be, what errors they will be asked to judge, and
how they will be asked to judge them. The judges can vary according to
whether they are native speakers (NS) or non-native speakers (NNS), and
also according to whether they are ‘expert’ (i.e. language teachers) or ‘non¬
expert’. The errors they have been asked to judge cover semantic or lexical as¬
pects of English, different grammatical features, and spelling. The instru¬
ments used to elicit judgements vary in a number of ways. In most cases they
consist of decontextualized lists of sentences containing either one or several
errors. These sentences are usually taken from actual samples of learner lan¬
guage (mainly written compositions) but they are sometimes contrived. In
some studies the sentences are contextualized. The errors can be presented
orally but are usually presented in writing. The judges may be asked to evalu¬
ate the ‘comprehensibility’ of the sentences containing the errors, the ‘serious¬
ness’ or the ‘naturalness’ of errors, or the degree of ‘irritation’ they arouse.
Sometimes they may be asked to correct the errors and to give reasons for why
they judged some errors as especially problematic. In some studies the judges’
comprehension of the erroneous sentences is also tested.
Error evaluation studies have addressed three main research questions: (1)
Are some errors judged to be more problematic than others? (2) Are there dif¬
ferences in the evaluations made by NS and NNS? and (3) What criteria do
judges use in evaluating learners’ errors? We will briefly consider the main
findings on each of these issues.
NS judges tend to judge lexical errors as more serious than grammatical
errors (for example, Burt 1975; Tomiyana 1980; Khalil 1985). They also tend
to judge global grammatical errors as more likely to interfere with compre¬
hension than local errors, although as Santos (1987) points out, there have
been conflicting results on this point. Burt defines global errors as errors that
64 The description of learner language

© © ©
i—
> CO
c n ^
c V-
o
© ©== © © © X © CO ®
© c © DO)- £ © © © © CO
C JD ©0-0 2 © O) o O >
©0-0
^
o ©c
41 O 0 O) £ o s 8 c ^ o o E c s S
© "2 0 T~ -P © — v. 0 U) © © > O) tl ® c 0
c ©
> LL © © >^ o © 8 8 © CZ*T 2 ©
O 4=
m <2
® S o o
P
c
o « ra °
U ” m 1.
® S £ ®
T3
©
8 o II E a
U) c o t T3© ©©
© © © ZZ-

§ 8
© o
2 8
O ^ i e 5
> CO
© p to ©
o ~ '0
'© o © C
n o) _3 o
o

-Q ©
o
c c
o J^.Q a>n C 0 O ©
o 8 o 5 ai
O ©
b£
«§ w - c © c

§1
— 0
© 0
£ -Q
o
© o © CO D © 3
X -*-> t5
13 •a E
Ora ■*= <=
ii
- © |e
c c
© 8
© 3
© © ©
.
'8 2
© o ®
©
_z ©
Coo
o ra i_
T3 D ■£ -a 2
o p LU ©

>. —
2
■o O o |S8
3 m r~ O 0 >-2 8 si 8 © O .■ti
© ° O *o <£ ® oo ro © © O) *c — ©
a> £ © © c o © LU © © 4= "O © c -n —
DC
i2 D>
© 4= m ©
■•&§£« > E o - 1o © © CL
fi
3 © © ^ © btL -a U)
o
^tz
8 i: > © © ©
® o tS
- t: 2 ©
e 8 «
-0 © ©
o 2 © 5 (/)</)-
ra ra § CO £ E co ^ •8 o> ©'= P© O
© © 0 © E 0 c m 0)
OZr ©
© 8°
_C *“ O © 5= kE? 3 8
o b © o o t <D 4= 41 O
z E E Z <:
ir -o
C 3 .E © °-
2 CO £ © I— © © Q. 3 © © o © © © Q- © O I- T3 CD CO CL CL C ©

© © CD - 3 ©
a) *; ■o ^ o £ o n >

05 £ C
© ._!_
■o b © U)
T3 2 © © O o "ra 8
2 £•
.3. 8
© -O
£ ©
•E ©
2 -c © ^ -n ©
oi£ © 05 © ^ o c
g- C © -E © ® ® ->
~ T3 i2 © © ■o > ■o ,u
_ n ■£ -O
■D u o © > * ■o 0
® O o 3 a O 0 © o © ® ® - © © ©
0 © L— jx b © CL O) U)
0) -S' £ ■o q.
O *fc
© "O
© o
© 4=
m -E S
© E°.®
©
© ©
_ o -Q
3
(0 = .O © o © © ©
o lo m 8 E ® -A © u © O 0 "O 3 ® 5
®'» • C ,-L
JX ^
8i” ©
1 8
- -s «
®
® E
o !*
> c > © ~o |S
— c
0 © c © 2 c
5
c
® ©
©
© ©
5 CO ® o. ”
3 © 4= ^
> © -TJ
ll w 2 iu 12 © 2 I “ CD C -C -g © 4r 2 8 ©
■D 8 2 ~ O © o O) o O) ® © S O £ c w
©
O .® Q- 8 ■%~0 «-8 « 0 © 2 8
■2.S-
Q. Q. O Q.
E 8 2 E
.® 8 © .9
O ■Q E t 4D © ”1-0 43 © Dr©
=3 0 co©
-Q
3
© -Q
© c
— — rn u
,3 o o o o o o © © 0 tr\
©
—* rr\
Woo w 2 Ro | CO CL © © CL CO CL © o as c o CO 3 > © ©

©
8 2 ■o 2
o -*© r*8 c
©
o
fc
c
O <D ©
U1
© ©
b © s_ n= © . ©
c
o
.1 2 O) ©
© o © os — 2
2 c © “ «« o ©
3 41 P g> l|t
© B o>.«2 © O © © ° % \ Q. ©
c c © © © © v -r © O =
■O o © T3 ©
© ■O C O © t
© © o
© 8 fc_ © - O r- o > Q- o c
© © ©
X Q
©
©
O) S I
>
®
c
| E
© 8 ©
° © -n
o ©
© 0 —
boo © O 3 « o c © © P ©
—
C
©
o
CO o c t o-g O) ~ o 8 .£ o ra = O co © ©
© o © ©
© © O) E ©
© S C ^ '1 8 S' ” -c © © © © ©
3 2 .9 ,<o c 0 0 E
CO
© 1 © .9- :\j © © 2 8 © o c
■£ o 3 o E = © c CL 8 ra E 8
©
5 Q_ co E
II <- © ©
CO £: O > U
o o o ,
CM S
.£ o
©
©
o E o.
t- < co m
I o t
ra viz 0

. 4= - CO
0).<2
© .. >> ©
©rc © .it ii
c © 0
© C0
CO o
o © £
©
4Z Q_
Z CO '© ^5 © > © 0
o c © .
3 c
■o =; 31 a> o
© —' c c
t.E . cl ^ o
CO © T-
z $ © © ©
CL ©
tr x © © If
CO -S
©
CL
<P -£ E ” ra
© ©
2 8
41
■n 05 © O
o X © § ■B & . © ©
© ©
ill
S
3
O) O)
C TJ © 6 ®
sll w 8 r- c Q. O CO
CO •»- 41 © CM 3 CO T— Z

C\1
~0 CO
c O)
©
c 3
© o
©
©
c o
© O
©
X ©
CO —I
c o cx> CO
CJ> ~o
c
in II ©
6- c
o> ©
■O
r
11 ©
3
■4-> 3 E © O)
I « o 4= 3
<0 CD < U- O X
 Learner errors and error analysis 65

VJ "o 0
0
Z c
Z © (7) 0
CO __
73 0
0 0
O
o ra
0
P 0
ro 2
w o 0 Bo 0 0
CD —
c ^
V-
0 :=
E to ^
3 0 — o
0
5 2Z c -O
■O 0 0 4-
M
p5
3 ^
= ® U)
BOO
P?
o
03
^ O)
z o/ 3 iz u
03
£ E Q. Q.^
•
si 0 CD 8 g i ®2 o
c ; -Q
© -c co § O
£ Q.
a?
si
0 0 0
£- O
5,8 o
P
0
0
O c
_ 0
73 J=
^ 3
«o2
0 = g?’g
>-. 0 2
o CO 0 0
0 7 0 ■o
£ gw
a -z
*s © 3
tu
0 £ "O
0‘S tl 2Z
+-
o
w g CL
§75

^
0 E 0 O P
Q. m-
0 O o "2
o
0
o P -9
0 _0
III 2o
0 Z
0
03
3 ~ 0
0 0 J=
E
£ 8 ®
2 ra f
O <-
11So
O 0
o

3
3
0
3 0
O .2
c p>
E 2
0
2 5s
b -Q 0
® g> g
l|| ■a 3
0 0
, 0 i2
2 42
c 2 g
ra 2 t 0 0 0 0
0
(/) 03-3 o o c
0
cc
2 "2 ° p 3 0 0 gdl q>
■O </> A
0=8 | § E
0 0 o
c '
^ o = 5 g>
0
IB |
6 « o
-
C -C
— 0
u/ Sa 13 CO ~T 0 .E Q. || o O) != ® 0 ~ 0
c 0 0
'(5 c
CO
>s
§5
0
C to
fs E
0
0
0
E
O
E c c
ra o "o
co o o « « g
Z ^3 X o
2 ra 4«: fc o 0
o £ > E 2 0 o
c b
2 z > © 0 W ro CO P O ojO x> Z 0 © 1 ® cr E z50 3 0

0 0 o 0
O) 03 0 03 -x O
a? c
73 0 n "O o
ra ® 3 © 0 3 0 0
*- to -P 0 .1® | 0 *- o ■O ~
o
2 2 2 c O 0 c b
0 •“
-a 2 ■a 2 S 0 TJ 0
®
-*
<E
®
0 3
-X 0 S 0 >■ 3 0
-X 2
r> ■g 8 O

A summary of selected error evaluation studies

~ 0 0
ra 0 0 m c 0 ^ 0
0 »♦_ 0 *0 5 w 0 0 ^8
0 0
J? £! to 0 o 0 ® 03
2 o ■£ 0 3 0 O 0 0 O 0
"v 0 0 ±=
ra « ® 0 X-? 0 0 0 O 0 ■o
1 4=- • ^ -o CL 0 0
3
*D
0
" E
_ 0
_ c E
o <« o
IS 2 5 8
® 2> g 73 iC
0 O)
^ 0
0
0 c 0 E 0 JZ
O 3
o 0 II 0
"O
0
.

O 0 0 8 1 2, J= 0 0
O
|ou
■Q-cTJ
0 *0
ft 2 a> c
3-^0
n o *o
3
-03 8
o
2
CO 0 0 CO 0 CO .E 0 CO 0 CO 0 3 0 0

CO _
c
0 o O CM g 03 . ■O © 03 *0
c b o ■»- O
>,g «
E '0 < 0 3 0
3 0-3 o CM "O B 3 ^ 0
O _ (J ? £ 0— -O 0
CO 0 0 r; -*-> -O o c C is:
0
0 c 0 0 0 co c .E C"O C .3
03 r~ c
co ~ cT' 3 0 ^ 0 3 •• E 03 0 _Q -- 0 0 0 0 P 0
o W o
® 9; <5 3 CM p 0 E
0 ^
w
±3 C 0
C -X
0 c >* ■* . 73
2 T3 0
P
P
b 0 > 3 O "5 I
0 '3 ^ 0 -n 03 0
»?o 0 0 r»
P -o c ® O 0 ^ >0 0
-o > >+- • •3 0
0 c 0
0 0 W U
0
0
P O)
CD 0
.
22 «
t «-o
O m
0
0 >
0 ^ > 8. o ■Q S
E S2 2 6 o
"O "c E 0 O iS O 0 N 0 O S§
0
0 A 0 O E CD j- 0
E 0 iz= P 0 O ■« 8 2 ® v
3 0 3 O >3 0 "O 0
a> I 8 8 0 O O 3
"* *n ® S
Ib © 0 g-g 0 C 0 0 ^
0 LO -* X
3
</> o
" 8
^ P
s s#
0 - I
i- s - 2 x
0 c o 2 Ia o c
CL 0 0 C
c\T 0
0 .E
(0 0 3 .
a)
o
“1
03 E a) 2 c
0 ^ E n
0 0
E 3
3 !l E «
_ -3 E c
5
C\J O
CO o
H—
o cm E 8.c
O
CM 0 2< 8 O 0
CM O
o
O 0
C V4
3 O 0 3©

0
0 CO
Table 2.3:

CO z 3
3
3 73 z
o 2 *4— 0
0
0 -£ o ^r
CO O
0 0
0 s L
0 0 ■o 0
n -o 3 © 0
C 0 0 O 0
E 3
0 2 0 Q. c
C 0 .3
E 0 .3 O ^0
§« >3 . O O 0
'l. 0 0 0 0

o
o ^
2sz 3
O 0
& 0 0 ^
2 0
E
a> O 0 45 0 < w g F
!n 2'a CD .> Z ‘‘O § &
3 i- C 3 O -3
co 5 iS CO 3 CM Z CO ^ CL

^r
co CO
03 CO 6-
CO 03 00
LO
CO
03 ■0 CO T_ 03
03
0 0 0
0 O
0 c O
> c 0 0 3
0 0 .3 .3 0
Q > CO CO
66 The description of learner language

affect overall sentence organization. Examples are wrong word order, miss¬
ing or wrongly placed sentence connectors, and syntactic overgeneraliza¬
tions. Local errors are errors that affect single elements in a sentence (for
example, errors in morphology or grammatical functors). NS judges may also
be influenced by markedness factors. Santos’ study, for example, lends some
credence to the idea that errors involving the substitution of marked for
unmarked forms (for example, ‘an book’ for ‘a book’) are judged more
severely than errors in which unmarked forms replace marked forms (for
example, ‘a apple’ for ‘an apple’). NS judges also find it easier to deal with
insertion than with omission or wrong choice errors (Tomiyana 1980).
It should be noted, though, that there can be considerable variation in
the judgements of native speakers. Thus Vann, Meyer, and Lorenz (1984)
found that some academic faculty members were inclined to view all
errors as equally serious—‘an error is an error’. Also, attempts to identify a
hierarchy of errors according to their effect on intelligibility have not proved
successful.
There are clear differences in the judgements made by NS and NNS. Over¬
all, NNS are much more severe (James 1977; Hughes and Lascaratou 1982;
Davies 1983; Sheorey 1986). NNS judges seem to be especially hard on mor¬
phological and functor errors in comparison to NS judges. However, they
tend to evaluate lexical and global errors less severely than NS judges.
Judges appear to use different criteria in assessing error gravity. Khalil
(1985) identifies three general criteria: intelligibility, acceptability, and irrita¬
tion. Intelligibility concerns the extent to which sentences containing differ¬
ent kinds of error can be comprehended. Acceptability is a rather vague
criterion, involving judgements of the seriousness of an error. Irritation con¬
cerns the emotional response of an addressee but is also related to the fre¬
quency of errors. Albrechtsen, Henriksen, and Frerch (1980), who had NS
judges rate the errors made by Danish learners of English in oral interviews,
found that ‘all errors are equally irritating ... irritation is directly predictable
front the number of errors regardless of the error type or other linguistic
aspects’ (1980: 394).
Obviously NS and NNS judges vary in the criteria they use. NS judges ap¬
pear to be more concerned with the effect that an error has on their compre¬
hension, whereas NNS judges are more influenced by their ideas of what
constitute the ‘basic’ rules of the target language (Hughes and Lascaratou
1982). However, Davies (1983) points out that NNS judgements will be in¬
fluenced by a number of factors relating to the particular context in which
they operate. Thus NNS teachers will be influenced by their background
knowledge of the syllabus and text book the learners are following and by ex¬
plicit knowledge of their LI. Transfer errors are viewed leniently, but errors
in grammatical structures that have already been taught will be seen as more
serious. Davies makes the general point that ‘any evaluation will be coloured
 Learner errors and error analysis 67

by the particular viewpoint from which it was carried out, and this may not be
consistent with evaluations made from other viewpoints’ (1983: 310).
As we noted in the introduction to this chapter, EA studies have often been
pedagogically motivated. They have sought to identify criteria for estab¬
lishing error gravity so that teachers can be guided in what errors to pay more
attention to. The general conclusion is that teachers should attend most care¬
fully to errors that interfere with communication (i.e. semantic and global
grammatical errors).9 Johansson (1973) suggests that errors should be evalu¬
ated by first asking whether they are comprehensible, and second whether
they cause irritation. Other, secondary factors—the frequency and generality
of the feature involved—also need to be considered. In this way, Johansson
constructs a hierarchy of errors. However, he acknowledges that ‘it is not
possible to illustrate the scale of errors at the present time since there is no
available information concerning the degree of comprehensibility/irritation
caused by different errors’ (1973:109). Although there has been considerable
research since, there is still insufficient evidence to support a definite scale for
evaluating errors.
Like other aspects of EA, the evaluation of learner error poses a number of
problems. It is not at all clear what criteria judges use when asked to assess the
‘seriousness’, ‘intelligibility’, or ‘acceptability’ of an error. As we have seen,
error evaluation is influenced by the context in which the errors occurred.
Thus, the same error may be evaluated very differently depending on who
made it and where, when and how it was made. The experimental studies
which have been conducted to date, however, take no account of these con¬
textual factors, often presenting errors for evaluation in isolated sentences. It
is perhaps not surprising that these studies have produced conflicting results
(see Santos 1987). The appearance of rigour given by the use of descriptive
and qualitative statistics may therefore be spurious.

The limitations of error analysis

There have been a number of critiques of EA (Bell 1974; Schachter and Celce-
Murcia 1977; Long and Sato 1984; Van Els et al. 1984). Bell goes so far as to
call EA a ‘pseudo procedure’. The criticisms levelled at EA fall into two main
categories: (1) weaknesses in methodological procedures, and (2) limitations
in scope. The first type have already been considered in the previous sections,
so we focus on the second here.
A frequently mentioned limitation is that EA fails to provide a complete
picture of learner language. We need to know what learners do correctly as
well as what they do wrongly. This problem has been overstated, however.
First, Corder (1971b) explicitly recognized the importance of examining the
totality of the learner’s production. Second, there is nothing to prevent the re¬
searcher doing this. At the very least, EA can be considered to have a place ‘as
68 The description of learner language

a partial and preliminary source of information at an initial stage of investiga¬

tion (Hammarberg 1973: 34). Indeed, EA continues to be used as a means of
investigating learner language with due attention paid to non-errors as well as
to errors.
EA is limited in a second way. Most of the studies are cross-sectional in na¬
ture, affording only a very static view of L2 acquisition. In many cases little
care has been taken to separate out the errors made by learners at different
stages of development. As a result, EA has not proved very effective in helping
us understand how learners develop a knowledge of an L2 over time. This
weakness is, again, not a necessary one. EA can be used in longitudinal studies
of L2 learners as in the Chamot (1978,1979) study referred to earlier. A study
of how learners’ errors change from one stage to another can shed light on the
process of L2 acquisition.
The third problem is more substantive, however. Schachter (1974) con¬
ducted an analysis of the relative clause errors produced by two sets of learn¬
ers (one Arabic and Iranian, and the other Chinese and Japanese). She found
that the first group of learners made more errors than the second group, des¬
pite the fact that relative clause structures existed in their Lis and did not ex¬
ist in Japanese and Chinese. However, she also discovered that the Arabic and
Iranian learners made many more attempts to use relative clauses than did the
Japanese and Chinese learners. She concluded that learners may resort to
avoidance if they find a structure difficult. Subsequent studies by Kleinmann
(1978), Kellerman (1977), Dagut and Laufer (1985), and Hulstijn and
Marchena (1989) testify to the prevalence of avoidance in L2 acquisition. EA,
which focuses exclusively on what learners do, has no way of investigating
avoidance and is, therefore, seriously limited. Avoidance is clearly an import¬
ant issue for SLA research; it is examined more fully in Chapter 8.

Summary

Error analysis (EA) was one of the first methods used to investigate learner
language. It achieved considerable popularity in the 1970s, replacing con¬
trastive analysis. The first step in carrying out an EA was to collect a massive,
specific, or incidental sample of learner language. The sample could consist of
natural language use or be elicited either clinically or experimentally. It could
also be collected cross-sectionally or longitudinally. The second stage in¬
volved identifying the errors in the sample. Corder distinguished errors of
competence from mistakes in performance and argued that EA should invest¬
igate only errors. Corder also proposed a procedure for identifying errors by
reference to normal, authoritative, and plausible interpretations. The third
stage consisted of description. Two types of descriptive taxonomies have
been used: linguistic and surface strategy. The former provides an indication
of the number and proportion of errors in either different levels of language
(i.e. lexis, morphology, and syntax) or in specific grammatical categories (for
 Learner errors and error analysis 69

example, articles, prepositions, or word order). The latter classifies errors ac¬
cording to whether they involve omissions, additions, misinformations, or
misorderings. The fourth stage involves an attempt to explain the errors psy-
cholinguistically. Competence errors can result from transfer, intralingual, or
unique processes. They can also be induced through instruction. EA studies
have produced widely differing results regarding the proportion of errors that
are the result of LI transfer, but most studies concur that the majority of
errors are intralingual. The precise proportion varies as a product of such fac¬
tors as the learners’ level, the type of language sampled, the language level (for
example, lexis v. grammar) and the learners’ ages. Also, errors can have more
than one cause. Finally, evaluation studies entail establishing the effect that
different errors have on the person addressed—either in terms of comprehen¬
sion or affective response. They have produced evidence to show that global
errors affect comprehension more than local errors, that non-native speakers
are inclined to be harsher judges of errors than native speakers, and that dif¬
ferent criteria involving intelligibility, acceptability, and irritation are used to
make judgements.
EA has lost popularity as a result of its perceived weaknesses. These weak¬
nesses include methodological problems involving all stages of analysis and,
also, limitations in the scope of EA. Focusing solely on the errors which learn¬
ers produce at a single point in time—as most of the studies have done—can
only provide a partial picture. It takes no account of what learners do cor¬
rectly, of development over time, and of avoidance phenomena.

Conclusion: a reassessment of EA
EA constituted the first serious attempt to investigate learner language in or¬
der to discover how learners acquire an L2. Its heyday was in the 1960s and
1970s. It then went out of fashion, as a result of the perceived weaknesses in
procedure and scope discussed above. Currently, it is showing signs of
making a come-back (see, for example, Taylor 1986 and Lennon 1991).
As we have already noted, some of the weaknesses are not inherent in EA.
Taylor (1986) outlines a number of principles that he believes should guide
the practice of EA. These principles are based on the general claims that ‘what
constitutes significant error is not strictly quantifiable’ and that we should
‘conceive our analytical aims to lie rather more in the interpretative traditions
of a humanistic discipline than has recently been customary’ (1986: 162).
Taylor demonstrates, through the detailed analysis of a piece of writing pro¬
duced by a native speaker, how the study of errors should be located in the
‘whole text’ and how it can afford valuable insights into the process of
language use. Lennon (1991) remains more committed to the quantification
of errors and seeks to show how some of the problems of error identification
can be overcome. He points out that most ‘erroneous forms are, in fact, in
themselves not erroneous at all, but become erroneous only in the context of
70 The description of learner language

the larger linguistic unit in which they occur’ (1991: 189). To take account of
this in error identification he proposes two new dimensions of error: domain
and extent. Domain refers to the breadth of the context (word, phrase, clause,
previous sentence, or discourse) which needs to be considered for de¬
termining whether an error has occurred. Extent refers to the size of the unit
(morpheme, word, phrase, clause, sentence) that requires deleting, replacing,
reordering, or supplying in order to repair an erroneous production. For ex¬
ample, in an error like *‘a scissors’, the domain is the phrase and the extent is
the word, while in an error like *‘well, it’s a great hurry around’, both domain
and extent are the whole sentence.10 Lennon illustrates how the concepts of
domain and extent can help to distinguish different kinds of lexical error. The
qualitative approach that Taylor recommends and the improved quantitative
approach proposed by Lennon have much to offer SLA.
EA, in fact, continues to be practised, although now it is more likely to
serve as a means for investigating a specific research question rather than for
providing a comprehensive account of learners’ idiosyncratic forms. For ex¬
ample, Felix (1981) and Pavesi (1986) use error analysis to compare the lan¬
guage produced by instructed and naturalistic learners. Bardovi-Harlig and
Bofman (1989) wished to investigate the differences between a group of
learners who successfully passed the Indiana University Placement exam and
a group who failed to do so. They examined the nature of the errors which the
two groups produced in one part of the examination—written composi¬
tions—and found, unremarkably, that the pass group made fewer overall
errors than the non-pass group and, more interestingly, that the major differ¬
ences were in the number of lexical and morphological rather than syntactical
errors. Santos (1987), in the study already referred to, carried out an error
evaluation in order to investigate linguistic claims regarding markedness.
Clearly EA is still alive, often used alongside other analytical techniques. It is
interesting to note, however, that in these recent studies there is no mention of
any of the methodological problems involved in EA.
EA has made a substantial contribution to SLA research. It served as a tool
for providing empirical evidence for the behaviourist/mentalist debates of the
1970s, showing that many of the errors that learners make cannot be put
down to interference. It helped, therefore, to support the claims made by Cor-
der, Dulay and Burt, and others regarding the ‘creativeness’ of much learner
language. Perhaps, above all, it helped to make errors respectable—to force
recognition that errors were not something to be avoided but were an inevit¬
able feature of the learning process. Indeed, the very concept of ‘error’ came
to be challenged on the grounds that learners act systematically in accordance
with the mental grammars they have constructed and that their utterances are
well-formed in terms of these grammars. As Corder put it, ‘everything the
learner utters is by definition a grammatical utterance in his dialect’
(1971a: 32).
 Learner errors and error analysis 71

Notes

1 This example of a sentence containing an error—and others in this chap¬

ter—is intended to reflect the kind of errors that are often made. In cases
where attested sentences are used to illustrate a point their source will be
indicated.
2 The study of production errors in native speakers has received explicit at¬
tention by Fromkin (1971), among others. As Crookes (1991) points out,
speech errors serve as the primary data for the construction of models of
language production. In retrospect, it is somewhat surprising that the L2
researchers who developed and made use of error analysis techniques
made so little reference to work on LI production.
3 Contrastive analysis is considered in greater detail in Chapter 9, where we
examine the role of language transfer in L2 acquisition.
4 Although most studies have employed either specific or incidental
samples, this need not be seen as a limiting factor, as it is possible to com¬
pare and aggregate the results of relatively small-scale studies in order to
construct a general picture. Richards (1971b) did just that.
5 Learners’ ‘model preferences’ (Beebe 1985) are a reflection of their refer¬
ence group, a point discussed in Chapter 6.
6 Some errors are likely to be easier to identify than others. Pragmatic
‘errors’, for example, are particularly difficult, partly because of the prob¬
lem of establishing which target language norms should serve as the point
of reference but mainly because it is essential to know what the learners’
exact meaning intentions were.
7 The distinction between competence and performance in an L2, which
here is discussed in terms of ‘errors’ and ‘mistakes’, has more recently
been considered in terms of the general distinction between ‘knowledge’
and ‘control’ (for example, Bialystok and Sharwood Smith 1985). This
distinction is discussed in the Introduction to Section 4.
8 Svartvik (1973b) refers to error evaluation as ‘therapy’.
9 The claim that greater attention should be paid to errors that affect com¬
prehension does not mean that no attention should be paid to local errors
in morphology or the use of grammatical functors. Indeed, the fact that
such structures do not have much communicative value may be one
reason why they are often not acquired and why an instructional focus on
them may be necessary (see Chapter 10).
10 Lennon (1991: 192) claims that ‘for any given error, domain will be at
higher rank than or equal rank to extent, but never at a lower rank’.
72 The description of learner language

Further Reading
There have been a number of collections of articles on error analysis. One of
the earliest is:
J. Svartvik (ed.), Errata: Papers in Error Analysis (CWK Gleerup, 1973).
More accessible, and also one of the best, is:
J. Richards (ed.), Error Analysis (Longman, 1974). This contains key articles
by Corder and Richards.
Another useful collection is:
B. Robinett and J. Schachter (eds.), Second Language Learning (University of
Michigan Press, 1983). In addition to articles by Corder and Richards, this
contains Duskova’s study, Stenson’s article on induced errors, and Schachter
and Celce-Murcia’s critique of error analysis.
Finally, there is a collection of Corder’s papers published in 1981 by
Oxford University Press (Error Analysis and Interlanguage).
Dulay and Burt’s work on errors has been subject to heavy criticism but is
worthy of consideration. Chapter 7, ‘Errors’, of the following provides a
good overview:
FI. Dulay, N. Burt, and S. Krashen, Language Two (Oxford University Press,
1982).
There are numerous error evaluation studies. One of the most thought-
provoking is:
E. Davies, ‘Error evaluation: the importance of viewpoint.’ English Language
Teaching Journal (19 83) 37: 304-11.
Finally, the following are strongly recommended for those who are think¬
ing of carrying out an error analysis of their own:
G. Taylor, ‘Errors and explanations.’ Applied Linguistics (1986) 7: 144-66;
P. Lennon, ‘Error: some problems of definition, identification, and distinc¬
tion.’ Applied Linguistics (1991) 12: 180-95).
3 Developmental patterns: order and
sequence in second language
acquisition

Introduction

The previous chapter began the study of learner language by focusing on

error analysis. A number of methodological and theoretical problems with
this type of investigation were identified. In particular, it was noted that error
analysis did not provide a complete picture of how learners acquire an L2 be¬
cause it described learner language as a collection of errors. Increasingly, re¬
searchers have recognized the need to consider the entirety of learner
language in order to uncover the systems of rules or interlanguages that learn¬
ers construct at different stages of development. Central to this enterprise is
the description of how learner language develops over time.
One of the most powerful ideas to have emerged from this work was that
L2 acquisition proceeds in a regular, systematic fashion. In this chapter we
will examine the descriptive evidence that lends support to the existence of
regular developmental patterns in L2 acquisition, reserving theoretical and
pedagogical considerations until later.
In this chapter, a distinction will be made between the idea of an order and
a sequence of acquisition. One question we can ask is ‘Do learners acquire
some target-language (TL) features before others?’ This is a question about
the order of acquisition. We can answer it by showing that one feature, say
plural -s in English, is acquired before another. A second and entirely differ¬
ent question is ‘How do learners acquire a particular TL linguistic feature?’
To answer this question we need to investigate some specific feature (such as
negation) in detail and, preferably, over time, in order to show how learners
gradually arrive at the TL. Showing that learners pass through stages on route
to the TL rule provides evidence for a sequence of acquisition. Developmental
pattern will be used as a cover term for the general regularities evident in lan¬
guage acquisition. As such, it subsumes the ideas of order and sequence.
The starting point for this chapter will be an account of the main methods
that have been used to study developmental patterns in language acquisition.
A brief examination of developmental patterns in LI acquisition follows, set¬
ting the scene for a more thorough account of L2 acquisition. The emphasis
will be on description (explanation being handled in later chapters) and on
74 The description of learner language

research carried out primarily with naturalistic L2 learners.1 The survey will
begin with a description of three aspects of early L2 acquisition: (1) the silent
period, (2) the use of formulas, and (3) structural/semantic simplification. It
continues with an account of research which has investigated the acquisition
of grammatical morphemes. There is a review of the morpheme studies,
which investigated whether the acquisition of a set of grammatical morph¬
emes (mainly in English) follows a fixed order, and there is also a detailed ac¬
count of acquisition in one grqup of morphemes (those belonging to the
pronoun system of a language) in order to show the stages of development
evident in these features. Next comes an examination of research which has
produced evidence to show that the acquisition of certain syntactic structures
follows a defined sequence, with learners manifesting various transitional
constructions in the process of acquiring TL rules. A brief discussion of the L2
= LI Hypothesis follows. Finally, there will be a general evaluation of the
claims regarding developmental patterns.

Methods for investigating developmental patterns

There are a number of different ways in which researchers can set about try¬
ing to identify developmental patterns. One way is to examine whether learn¬
ers’ errors change over time. There is some evidence to show that this does
happen, but as we saw in the last chapter, error analysis has not succeeded in
providing clear and conclusive evidence of developmental patterns. A second
way is to examine samples of learner language collected over a period of time
in order to identify when specific linguistic features emerge. According to this
approach, ‘acquisition’ is defined as ‘first occurrence’. It has been used extens¬
ively in first language acquisition research (for example, Wells 1985) but is
less common in SLA research. However, a number of L2 researchers (for ex¬
ample, Bickerton 1981 and Pienemann 1984) have recommended the use of
emergence as the criterion of acquisition in SLA research.
One common method for identifying and describing developmental pat¬
terns is obligatory occasion analysis. This has been widely used by L2 acquisi¬
tion researchers and is clearly described in Brown (1973). The basic
procedure is as follows. First, samples of naturally occurring learner language
are collected. Second, obligatory occasions for the use of specific TL features
are identified in the data. In the course of using the L2, learners produce utter¬
ances which create obligatory occasions for the use of specific target-language
features, although they may not always supply the features in question. Thus,
if a learner says: ‘My sister visited us yesterday’ or “-‘My father arrive yester¬
day’, obligatory occasions for the use of past -ed have been created in both
utterances. Third, the percentage of accurate use of the feature is then calcu¬
lated by establishing whether the feature in question has been supplied in all
the contexts in which it is required. A criterion level of accuracy can then be
determined in order to provide an operational definition of whether a
 Developmental patterns: order and sequence 75

feature has been ‘acquired’. Usually, the level is set at 80-90 per cent, below
100 per cent, to take account of the fact that even adult native speakers may
not achieve complete accuracy. Brown (1973) considered a feature to be
‘acquired’ if it was performed at the 90 per cent level on three consecutive
data collection points—a very rigorous definition.
One problem with obligatory occasion analysis is that it takes no account
of when a learner uses a feature in a context for which it is not obligatory in
the TL. For example, the learner who says ‘*1 studied last night and now I un¬
derstood better’ has overgeneralized the past tense, using it where the TL re¬
quires the present tense. Clearly, acquisition of a feature such as past tense
requires mastering not only when to use it but also when not to use it. To take
account of over-uses as well as misuses a number of researchers (for example,
Pica 1983) have suggested a procedure known as target-like use analysis. Pica
(1984) has shown that substantial differences in estimates of learners’ abil¬
ities arise depending on whether obligatory occasion or target-like use ana¬
lysis is employed.
Both obligatory occasion and target-like use analysis are target-language-
based—that is, like error analysis, they seek to compare learner language and
the TL. Bley-Vroman (1983) has pointed out the dangers of what he calls the
‘comparative fallacy’. The main one is that it ignores the fact that learners cre¬
ate their own unique rule systems in the process of learning an L2. Target-
language-based analyses cannot be used to describe these systems as they only
provide information about the extent to which the learner’s language approx¬
imates to the TL.2 One way of overcoming this objection, which some re¬
searchers adopt (for example, Cazden et al. 1975), is to catalogue the various
linguistic devices that learners use to express a particular grammatical struc¬
ture (such as negatives) and then to calculate the frequency with which each
device is used at different points in the learners’ development. This method
has no commonly recognized name but might usefully be called frequency
analysis. It is able to show the ‘vertical variation’ in learners’ development
(i.e. how different devices become prominent at different stages) and serves as
one of the best ways of examining developmental sequences.
Many of the studies considered in this chapter are longitudinal, involving
data collection over a period of many months and, in some cases (for ex¬
ample, Schmidt 1983) several years. Such studies provide the strongest evid¬
ence in support of developmental patterns. In addition, there were a number
of studies that were cross-sectional in design (i.e. data were collected only at a
single point in time). In order to make claims about the order of acquisition
on the basis of cross-sectional data, researchers resorted to a number of stat¬
istical procedures. For example, some researchers argued that the accuracy
order with which different features were performed corresponded to their ac¬
quisition order (for example, Dulay and Burt 1973 and 1974c). Thus, for ex¬
ample, if the data showed that short plural -s (as in ‘boys’) was performed
more accurately than long plural -es (as in ‘churches’), then this indicated that
76 The description of learner language

the short plural form was acquired before the long plural form. An alternative
procedure for establishing order of acquisition from cross-sectional data in¬
volves implicational scaling. This technique was first used in Creole studies
(Decamp 1971). It seeks to exploit the inter-learner variability that exists in a
corpus of learner language in order to establish which features different learn¬
ers have acquired and whether the features can be arranged into a hierarchy
according to whether the acquisition of one feature implies the acquisition of
one or more other features for each learner. For interested readers, Flatch and
Farhady (1982: chapter 14) provide a clear explanation of how implicational
scaling is carried out. An example can be found in Figure 4.1 on on page 126
of this book. The extent to which cross-sectional data, so processed, can pro¬
vide valid information about developmental sequences is a matter of some
dispute.
The existence of developmental patterns can be investigated in different
areas of language: linguistic (phonological, lexical, and grammatical),
semantic, and functional. This chapter will deal more or less exclusively with
the acquisition of grammatical systems—mainly because this reflects the gen¬
eral emphasis on grammar in SLA research, and serves as the primary evid¬
ence for theory construction. However, limited attention will also be given to
phonological and lexical aspects of learner language, while Chapter 5 will ex¬
amine the acquisition of pragmatic features.

Developmental patterns in LI acquisition

The study of developmental patterns in LI acquisition research is a good
starting point in our investigation of L2 developmental patterns for two
reasons. First, it has provided L2 researchers with useful methodological pro¬
cedures for investigating developmental patterns in learner language. Second,
LI acquisition orders and sequences provide a baseline for considering L2
acquisition orders and sequences. An important issue is whether the patterns
in the two types of acquisition are the same or different.
Early work on LI acquisition in the 1960s and 1970s consisted of detailed
case studies of individual learners based on the speech they produced (for ex¬
ample, McNeill 1970; Slobin 1970; Brown 1973), of cross-sectional studies
of larger numbers of learners (for example, de Villiers and de Villiers 1973),
and of some experimental studies of children’s production and comprehen¬
sion of specific linguistic features (for example, Berko 1958; C. Chomsky
1969). These early studies, which attempted to identify the general pattern of
children’s language development, began what Atkinson (1986) has referred
to as the SHARP tradition in first language acquisition research (i.e. the em¬
pirical study of LI acquisition).3 This tradition has continued into the 1980s
and 1990s, spawning an enormous amount of research in a great range of
languages. Notable recent projects include the Bristol Study, ‘Language at
Home and at School’ (Wells 1985), which involved the longitudinal study of
 Developmental patterns: order and sequence 77

more than 60 children and the enormous cross-linguistic project initiated by

Slobin (see Slobin 1985a). Inevitably in this span of time there have been
shifts in emphasis, but in this tradition researchers largely continue to follow
a research-then-theory approach, emphasizing the description of children’s
language as a basis for theory building. An alternative tradition, referred to as
FLAT (First Language Acquisition Theories) by Atkinson, grew up in the
1980s. This eschews empirical enquiry and instead seeks to examine LI
acquisition from the point of view of learnability theory and Universal
Grammar. In this chapter we will be concerned exclusively with the SHARP
tradition, although we will also briefly consider some of the theoretical issues
with which this tradition is concerned. Learnability theory and Universal
Grammar will be considered in detail in Chapter 10.
One of the pervasive findings in first language acquisition research is that
children appear to follow a fairly well-defined pattern of development. This
pattern is evident in the way in which all linguistic systems are acquired. Chil¬
dren typically begin with one-word utterances which function as holophrases
(i.e. express whole propositions). They gradually extend the length of their
utterances, passing through stages when the bulk of their speech consists of
first two-word, then three- and four-word utterances. At the same time, they
systematically acquire the various syntactical and morphological rules of the
language. The result is that remarkable regularities are evident in both the
overall pattern of development and in the acquisition of specific linguistic sys¬
tems (for example, tense markings or negatives). These regularities are often
described with reference to mean length of utterance as a general measure of
development. Crystal (1976), for example, describes the specific grammatical
features that are evident in children’s language when the mean length of their
utterances is one word, two words, three words, etc.
In one of the most influential studies of LI acquisition—the Harvard
Study—Brown (1973) found evidence for the fixed order of acquisition of
various English morphological features in three children. The same order was
obtained in de Villiers and de Villiers’ (1973) cross-sectional study of twenty
children. Table 3.1 gives the order. The acquisition of individual morphemes
also involves stages. For example, the acquisition of English past tense forms
involves an initial stage in which there is little or no use followed by sporadic
use of some irregular forms, then use of the regular -ed form including over¬
generalization to irregular verbs, and finally target-like use of regular and ir¬
regular forms. Thus, the acquisition of forms such as ‘went’ follows a
U-shaped pattern of development, with children first using it correctly (for ex¬
ample, ‘went’) and then incorrectly (for example, ‘goed’) before they finally
once again produce the correct form (‘went’).
Clear examples of developmental sequences in LI acquisition are those
found in the acquisition of English negatives and interrogatives. Klima and
Bellugi (1966), for example, identified three stages in the acquisition of negat¬
ives, which are summarized in Table 3.2. Bloom (1970) found evidence of the
78 The description of learner language

Morpheme Example

1 Present progressive -ing He is sitting down.

2 Preposition ‘in’ The mouse is in the box.
3 Preposition ‘on’ The book is on the table.
4 Plural -s The dogs ran away.
5 Past irregular The boy went home.
6 Possessive ~’s The girl’s dog is big.
7 Uncontractible copula ‘be’ Are they boys or girls?
8 Articles ‘a’/‘the’ He has a book.
9 Past regular -ed He jumped the stream.
10 Third person regular -s She runs very fast.
11 Third person irregular e.g. has/does Does the dog bark?
12 Uncontractible auxiliary ‘be’ Is he running?
13 Contractible copula ‘be’ That’s a spaniel.
14 Contractible auxiliary ‘be’ They’re running very slowly.

Table 3.1: Order of LI acquisition in English morphemes (abridged from Clark

and Clark 1977:345)

systematic acquisition of the semantic functions which negatives can realize

in English. The children she studied first used ‘no’ and ‘not’ to refer to non-
presence (for example, ‘No cookie’), then to refer to rejection of an offer or
suggestion (for example, ‘No car’ = ‘I don’t want the car’), and finally to
denial. Similar regularities are evident in the acquisition of the formal and
semantic aspects of English interrogatives (see Klima and Bellugi 1966 and
Cazden 1972).

Stage Description Examples

1 Negative utterances consist of a ‘nucleus’ (i.e. the Wear mitten no.

positive proposition) either preceded or followed by a Not a teddy bear.
negator.
2 Negators are now incorporated into the affirmative There no squirrels.
clauses. Negators at this stage include ‘don’t’ and You can’t dance.
‘can’t’, used as unitary items. Negative commands Don’t bite me yet.
appear.
3 Negators are now always incorporated into affirmative I don’t have a book.
clauses. The ‘auxiliary + not’ rule has been acquired, as Paul can’t have one.
‘don’t’, ‘can’t’, etc. are now analysed. But some I not crying.
mistakes still occur (e.g. copula ‘be’ is omitted from No one didn’t come.
negative utterances and double negatives occur),

Table 3.2: The LI acquisition of English negatives (examples from Klima and
Bellugi 1966)
 Developmental patterns: order and sequence 79

These regularities in grammatical development are the product of the ac¬

quisition task which the child faces. Clark and Clark (1977) describe this task
as follows:

From the outset children are faced with two general problems. First of all,
they have to figure out how to map their ideas and general knowledge onto
propositions4 ... Second, they have to find out how to communicate speech
acts and thematic information along with the propositional content of their
utterances (Clark and Clark 1977: 296).

They emphasize that the two tasks of mapping and communicating go hand
in hand. This claim is supported by the fact that regularities are also evident in
the way in which the different pragmatic and textual functions of the TL are
mastered. Thus, for example, Clark and Clark illustrate how initially children
learn to perform assertions and requests and only later develop the ability to
express directives (for example, asking, ordering, forbidding, and permitting)
and commissives (such as promising). Expressives (for example, thanking)
and declarations follow even later. A full account of the developmental path,
therefore, must describe how children master the formal, functional, and se¬
mantic properties of a language. A good example of such an account is that
provided by Wells (1986a) and summarized in Table 3.3. It is important to re¬
cognize, however, that although certain stages of acquisition can be ident¬
ified, development is, in fact, continuous. Children do not usually jump from
one stage to the next but rather progress gradually with the result that ‘new’
and ‘old’ patterns of language use exist side by side at any one point in time.
It is also important to acknowledge the inter-learner variability that exists
in LI acquisition. This is most evident in the rate of acquisition. Some chil¬
dren learn their LI with great rapidity while others do so much more slowly.
Wells (1986b), for instance, notes that the difference between the most and
least advanced children when they were three and a half years old was equi¬
valent to almost 36 months—an enormous discrepancy. It is for this reason
that it is not possible to describe the sequence of development in terms of age.
The individual differences go further than rate, however. They concern the
overall strategy that children appear to follow. Lor example, although many
children use an analytical strategy and show evidence of the developmental
progression described above, other children use a gestalt strategy, typically
remaining silent for a longer period before producing full sentences when
they first start talking (Peters 1977). All children make use of unanalysed
units (formulas) but some seem to rely on them much more extensively than
others (Nelson 1973). There is a considerable body of research into LI ac¬
quisition that has sought to identify the factors responsible for inter-learner
variation. The variables studied include sex, intelligence, personality and
learning style, social background, and experience of linguistic interaction
(Wells 1986b).
The results of the empirical studies of LI acquisition have been used to
80 The description of learner language

+
CO -Q ~o
c _ 0 ^
o0 O c 0
o ■
. «
c> 0
£ 5
0
2 -3 £ .2 + .
i 5 •-PO H-o
0 §
ZoO 2D --P 0 O9 to t> ^ <1) «
gl
cr CO c <B
0- + O O 1 sz V I) 0 1 c
-2 o
S’ O' O o «8 + c U 0
o o
S S^
0 co o d a) 2 m 0 0
2
0
0
6 . _0o
5 d)£ to 5 0 o
O
3 ® ~ O
£ -2 Q.
0

O -Q c -Q
t" ^ ■c
0
.g d) Q- o g C 0
c O O JZ 8.1 9 0 SZ
~G CO i5 > CL
& ® 0 O >® ®E 0 0 -E V o 0c
F CL
E O :I
> to CD
di, ij 0
.2 Z 0 5 0
F c
--
- H— I 0 3 ^E 0 = 0 E “ 3 ^ ^ • l! "o .2
^O I i= c & o 0C tZD I; O -T-J
p a
o 2
gt
0 !o II
3co .9» ®E?£O 0
<D +Z l rf vo
> .2 - CL § CD ^
i O
0 p .E 0 M- *0 • ^ o -Q = Co
W w 1
0O 9O -„ 2- c©
0 i2 s O ‘O C -2 13 0 0
tr 0 © ® ON
o 'C ® 8 — Z3 I?
- a
JD 0 sS te
o . hz >
0 0 0 0
>
1
TJ
c
C O <
i5 g
0^ 0 S ~o C 0 “o 9- F — >0 > Q O 0
0 ;
0 0 c o O 0 (0
0 <D O " ® 0 + C 0 -P 0 © a ®
o O Q. 0 oo ^ 0 -E §
S CO

.1 § c *♦— ._ a? 8
~ w-^ m o
Forms

0 0 o o 0 0 CL 0 0
|5
0 0
!q 0 CL
-= 0 ^ • 0 £ *0 0
0 0
8 P
;5 o 6 0
=3 CO £: z>
0 0
X 2
1 8
fig O
to 2 > s
Dr » O E =3 O) C CL
3 2 a£
s ©
O CL
=5 0
< O - O 0 cr
0 O ® £
“ c
o
^3

0 0
o
c
CD
0
0 ' * o 0 0
c
T—H
o 0 to - § ^ £ 0 • 0 0

■g s ^ro ®q Q.
a
0 c 0 0 _ >
Q)
(0
O
SZ
1
E © 0 F
13 Q-
3 ' CT3 0 ^
5. _ -SZ ~oc -o0 0_ "O o
=5 5
!|
o Q-
o OT o O o
® 0 Q.--
0 C 0 O
To ^ CD 10 0 >■ ~o .
8 § to g ® E _ 0 "O
CO c
•-P -O C °- “a ^ o. a> 3 o
c © E § co 0 •
O Q. c ro
0 0
© E oil c w
0 2.
w g 0 0 ©. c O CL O o m R
■o0 -2
C o CD 0 .
O c a CL 0 OT X 3= ~
CD c £—p 0 0 0 m ® u 0
0 if*
F 0 .> 0 C CD £.1 TD CL
S' °
o ° i o -D CD E Q. C —
0
© CD 0 9
0a) .y-
& -n 0 9
CL 0
o 0
C
C
0 >2- .2
CL
o 0
® ra
0
S
X E o) 2 0 > sr^ 0 <Lj
!« « I O c
0 0 - .2 ^ -Q o >
2 -2
0 ? 0
0 0 £ o e § 0 C
2 § 0 0 0 _
> -z= 0
~ § a
|i F 5
.E 0 Cl
=>s+3 o c ® C CL ■O ;C
CO ■ 0
0 -2
0 © 0 o "O ~ o a®
o 0 « ®. .2 0 o
2 CO
Meanings

C ©
I? ai’D £
0 fi o
9- CJ)
>? c: 0 1 g 8 'I
0 0
c C 0 _o 9 0
o 5 0 0 'SZ
CL
^3
-c it d L— n o
0 Own c E ® § o 0 0 0 2 - c CL
CL 0 0 -c 0 O 0 .2
Q. _c :£ -C 0 0 4-> 2 ^ 1 SZ ' 0 cbi
0
_2 i- o d) -
co 1—0 0 O 0 CL 0 0
o
CL

■oc CD
•S
c 0 +<.
.9 0 0
0 _0 s
S c 0“ ‘O
c0 ~Bo a 0 ^
0
O C 0
o
E
CNJ
co . _
oo 0 c ® g i "3
Oi.P
■oc § © ® CL
CD co 0 £ t
i- co FT a* cl ® R
g c
= co o CD 2-
£ 3 ■K ® Q. CL
co ^
0
E
0 0 “t c O
o =L 0 ra o o
^ CD
T-, C 0
~G
C 8 §
.b CD ■8 oi g
.. 0
0
■o0
SZ
T3 C 0^0
>> 0
= 0) > 0
^ 0 C
o 9 co
■B
■= x
Q- 0
sz
0 2 ~ C\J
o
0
=3 Q..2 CD
cr x: -E
.£ 0) !q 0
0
s
■g S If
1.P
cr
t- r O
t O) £
CD C
CO 0
0 C
0 ^ CO o © 2
O) © CL O c _E 11
CO c .Li O
O 0 0 0
8^|
CO O ®I 8 § O 'O >
Functions

CD C 0
9 CO
i.o
'To=- c
© 0 o 0 >
o co CLC/) co F
g to
0
lo
0 C
~ 0
£ -2
§ O 0 ^ S
_ o
^ C c^-
0) © _
0 >,
©
> 0 E
D Q- 3 J5
Q.^
12? ®
C\J
CO O 0 u_ a SiS5 z ^ 0
Stage
 Developmental patterns: order and sequence 81

evaluate different theoretical positions. Although our concern in this chapter

is primarily with the description of developmental patterns in learner lan¬
guage, it is helpful to understand the nature of these positions in broad out¬
line, particularly as they also figure in SLA research. The key debate in the
1960s and 1970s revolved around the rival claims of behaviourist and men-
talist models of acquisition. According to the former, children acquired their
LI by trying to imitate utterances produced by their parents and by receiving
negative or positive reinforcement of their attempts to do so. Language ac¬
quisition, therefore, was considered to be environmentally determined. Such
a model does not accord with the empirical facts, however. For example, chil¬
dren receive remarkably little formal feedback on the correctness of their ut¬
terances, many of which are uniquely created rather than imitated. Indeed,
the behaviourist model was advanced not on the basis of studies of child lan¬
guage but through extrapolation from studies of animals in laboratory
settings.
A mentalist model makes the following claims (see Ellis 1985a: 44):

1 Language is a human-specific faculty.

2 Language exists as an independent faculty in the human mind (i.e. it is sep¬
arate from the general cognitive mechanisms responsible for intellectual
development).
3 The primary determinant of acquisition is the child’s ‘language acquisition
device’, which is genetically endowed and provides the child with a general
set of principles about language which can be used to discover the rules of a
particular TL.
4 Input data are required to trigger the process by which the ‘language ac¬
quisition device’ discovers the rules of the TL.

This model is closely associated with the work of Chomsky (for example,
1965 and 1980a) in linguistics and Lenneberg (1967) in the biological pre¬
requisites of language. Precisely how the ‘language acquisition device’ works
is a matter of some controversy, however. Some LI acquisition researchers
(for example, McNeill 1966) have argued that children form hypotheses
about the rules to be found in the TL and then test them out against input data
and modify them accordingly. Other researchers (for example, Braine 1971)
have challenged this view on the grounds that hypothesis-testing is too ineffi¬
cient. They seek to explain acquisition in terms of discovery procedures
which are used to scan input data and to store noticed features.
Although much of the early discussion centred on the behaviourist/
mentalist debate, considerable attention has been given also to a third type of
model: a cognitive processing model. Proponents of such a model (for ex¬
ample, Sinclair-de-Zwart 1973) agree with the mentalists that children must
make use of innate knowledge, but disagree about its nature. Whereas
mentalists consider that it takes the form of a specific language faculty,
cognitive psychologists argue that it consists of a general learning mechanism
82 The description of learner language

responsible for all forms of cognitive development, not just language. They
point to the non-linguistic origins of language acquisition (i.e. the sensory
motor stage that precedes the onset of speech in children) and to the concur¬
rent development of linguistic and cognitive knowledge. One of the clearest
accounts of the kinds of cognitive processing involved in LI acquisition is to
be found in Slobin’s (1973 and 1985b) description of the operating principles
that children learning a variety of Lis resort to in order to learn from input.
These are discussed in some detail in Chapter 9.
The debate between the behaviourists and the mentalists was a poor con¬
test, easily won by the mentalists. However, the arguments regarding the na¬
ture of the internal mechanisms responsible for LI acquisition continue to
find current expression in the rival claims of cognitive and linguistic theories
of L2 acquisition (such as the Competition Model (MacWhinney and Bates
1989) and Universal Grammar (White 1989a)), discussions of which can be
found in Chapters 9 and 10 respectively.
In the sections that follow we will see a number of parallels between work
on developmental patterns in LI and L2 acquisition research.

Developmental patterns in second language acquisition

The research that we shall be reviewing has investigated learners in natural¬
istic settings and has been based on what might roughly be called ‘unplanned
language use’ (i.e. the learner language that results from attempts by learners
to express their meaning intentions more or less spontaneously). L2 learners,
particularly adults, have the capacity to engage in ‘planned language use’ by
paying deliberate attention to the language forms they choose (for example,
by using explicit knowledge of grammatical rules or by translating).
Unplanned and planned language use display markedly different features, as
we shall see in the next chapter. The idea of ‘developmental patterns’ is based
on unplanned language use and it is not at all clear to what extent it is applic¬
able to planned language use.

The early stages

The early stages of L2 acquisition in naturalistic settings are often charac¬

terized by a silent period, by the use of formulaic speech, and by structural
and semantic simplification.

The silent period

In the case of LI acquisition, children go through a lengthy period of listening

to people talk to them before they produce their first words. This silent period
is necessary, for the young child needs to discover what language is and what
it does. In the case of L2 acquisition, the silent period is not obligatory, as the
learner already knows about language, having already acquired one. Yet
 Developmental patterns: order and sequence 83

many learners—especially children—opt for a silent period. Itoh and Hatch

(1978), for instance, describe how their subject, Takahiro—a two-and-a-
half-year-old Japanese boy—refused to speak English at an American nur¬
sery school and also to the researcher in his own home for the first three
months. Hakuta (1976) reports the difficulty he had in obtaining any data
from his subject, Uguisu—a 5-year-old Japanese girl—for the first three
months, but then comments that her English suddenly ‘blossomed’. Saville-
Troike (1988) reports that six out of the nine children learning L2 English
that she studied opted for a silent period. Hanania and Gradman’s (1977)
study of the acquisition of English in the USA by Fatmah, a 19-year-old Saudi
woman, also indicates that some adults go through a silent period. Krashen
(1985) cites Rodriguez (1982) as another example of an adult learner who
began with an extensive silent period. Rodriguez describes his own case
history of learning L2 English, noting that he said nothing in class in the
American school he attended for the first six months.
Of course, not all learners go through a silent period, as Saville-Troike’s
study shows. Many learners—particularly classroom learners—are obliged
to speak from the beginning. But even when production is not required, some
learners opt for it. Paul, a 5-year-old Taiwanese boy studied by Huang and
Hatch (1978), appears to have begun talking in English almost immediately,
although most of his early utterances involved imitation. Studies of ‘good lan¬
guage learners’ report that many adult learners (82 per cent in a study carried
out by Naiman et al. 1978) claim they begin to speak right from the start.
Gibbons (1985) reviewed the evidence in favour of a silent period in both
children and adults and found it inconclusive. His own survey of 47 children
learning English as an L2 in Sydney primary schools revealed considerable
individual variation, with a mean length of just two weeks’ silence. It is also
interesting to note that in many of the studies which Krashen cites as provid¬
ing evidence of a silent period, the learners were not, in fact, completely silent,
but often produced some formulaic expressions right from the beginning.5
The question arises as to why some learners opt for a silent period while
others do not. Saville-Troike (198 8) suggests that the reason may lie in differ¬
ences in the learners’ social and cognitive orientation. She distinguishes
other-directed and inner-directed learners. The former ‘approach language as
an
interpersonal, social task, with a predominant focus on the message they wish
to convey’, while the latter ‘approach language learning as an intrapersonal
task, with a predominant focus on the language code’ (page 568). She sug¬
gests that while other-directed learners do not typically go through a silent
period, inner-directed learners do.
There is some disagreement regarding the contribution that the silent
period makes to language learning. Krashen (1982) argues that it provides an
opportunity for the learner to build up competence via listening. According to
this view, speaking ability emerges naturally after enough competence has
been developed through listening. Itoh and Hatch (1978), however, take a
84 The description of learner language

different view, referring to Takahiro’s silent period as a ‘rejection stage’, dur¬

ing which he tried to avoid learning English. Rodriguez (1982), cited by
Krashen 1982, claims that he only began to learn English when he started to
speak it at home with his parents. Gibbons (1985) concludes that the initial
silent period is in many cases a period of incomprehension that does little or
nothing to promote acquisition and that if the silent period is a prolonged one
it may reflect psychological withdrawal.
One possibility is that the silent period provides learners with opportun¬
ities to prepare themselves for social use of the L2 by means of private speech,
which they engage in while they are ‘silent’. Saville-Troike, in the study of
child L2 learners referred to above, defines silent speech as speech that is pro¬
duced at a very low volume so as to be inaudible to anyone present and with
no apparent expectation of a response. She used a radio-microphone hung
round the neck of the children to record it. Five of the children who went
through a silent period manifested private speech, using a variety of intra¬
personal strategies. These included repeating other speakers’ utterances,
recalling and practising English words and phrases, creating new linguistic
forms, substituting items in utterances, and expanding them and rehearsing
utterances for overt social performance. All five learners eventually began to
speak using first single words, memorized chunks, and repetitions of other
children’s L2 utterances. A sixth child who manifested a silent period, how¬
ever, did not engage in silent speech and, unlike the others, remained silent
throughout the study (approximately 18 weeks), apparently not learning any
English. Saville-Troike’s study suggests that while some child learners may
use silence as a strategy for avoiding learning, many make active use of it to
prepare for the time they begin speaking the L2. However, as Saville-Troike
acknowledges, her study does not show whether the learning strategies some
learners employ during their silent period are related to long-range L2
development.

Foripulaic speech

Formulaic speech consists of ‘expressions which are learnt as unanalysable

wholes and employed on particular occasions’ (Lyons 1968: 177). Hakuta
(1976) and Krashen and Scarcella (1978) distinguish routines and patterns to
refer respectively to whole utterances learnt as memorized chunks (for ex¬
ample, ‘I don’t know’) and to utterances that are only partially unanalysed
and have one or more open slots (for example, ‘Can I have a_?’). In Ellis
1984b, I too suggested that formulaic speech can consist of entire scripts,
such as greeting sequences, which the learner can memorize because they are
fixed and predictable.
Formulaic speech can be observed in the speech of native speakers.
Nattinger and DeCarrico (1992:1) suggest that what they call ‘lexical
phrases’ (defined as ‘multi-word lexical phenomena that exist somewhere
 Developmental patterns: order and sequence 85

between the traditional poles of lexicon and syntax’) are commonly used by
native speakers, reflecting the ritualization of language behaviour. Pawley
and Syder (1983) argue that achieving native-like control involves not only
learning a rule system that will generate an infinite number of sentences, but
also ‘memorized sequences’ and ‘lexicalized sentence stems’. They state that
‘the number of memorized complete clauses and sentences known to the ma¬
ture English speaker is probably many thousands’ (1983: 205). Some ex¬
amples are ‘Can I come in?’, ‘What’s for dinner?’ and ‘Speak for yourself’.
Sequences longer than simple, single clauses also exist. In addition, native
speakers internalize sentence stems in which the structure is fully specified to¬
gether with a nucleus of lexical and grammatical morphemes. An example is:

NP bt-tense sorry to keep-tense you waiting

which can be realized in a number of forms:

I’m sorry to keep you waiting.

I’m so sorry to have kept you waiting.
Mr X was sorry to keep you waiting the other day.

Pawley and Syder characterize the task facing the language learner as to dis¬
cover precisely what permutations of a sentence stem are possible—a view of
learning which bears a strong resemblance to that which underlies some cur¬
rent models of language use and acquisition, such as Parallel Distributed Pro¬
cessing (see Chapter 9). Formulaic expressions frequently embody the
societal knowledge which a given speech community shares and, according to
Coulmas, ‘are essential in the handling of day-to-day situations’ (1981: 4).
They enable the speaker to say the right thing at the right time in the right
place. They also perform a psycholinguistic function in speech production.
Dechert (1983) suggests that fixed expressions serve as ‘islands of reliability’
that help speakers to construct and execute production plans, a point ex¬
plored more fully in Chapter 9.
Formulaic speech has been observed to be very common in L2 acquisition,
particularly in the early stages. It figures frequently in the speech of all learn¬
ers, irrespective of their age. Itoh and Hatch note two very common patterns
in the speech of Takahiro, once he began talking—‘I get...’ and ‘I wanna ...’.
Ervin-Tripp (1974) notes that one of the first utterances produced by young
English-speaking learners of L2 French was ‘Peut-je jouer avec Corinne?’ and
notes that ‘the size of the unit stored is impressive’. Hakuta (1986) discusses
three patterns in the speech of his subject, Uguisu—those using copula, ‘do
you’ in questions, and ‘how to’ in embedded ‘how’ questions. Rescorla and
Okuda (1987) report similar patterns in their subject, Atsuko, a six-year-old
Japanese child. Hanania and Gradman (1977) observe that at the start of
their study, Fatmah relied almost entirely on ‘memorized items’. They note
that ‘the use of these expressions does not imply that she recognized the indi¬
vidual words within them, or that she was able to use the words in new
86 The description of learner language

combinations’ (page 78). Initially, Fatmah seemed to resist segmentation of

expressions such as ‘Thank you, I can’t...’ and ‘Do you like ... ?’ There are,
in fact, few case studies based on naturally occurring learner language that do
not make some mention of the prevalence of formulas.
What sets these formulas off from other samples of learner language is their
well-formedness. Thus, routines and patterns typically manifest TL morpho¬
logy and syntax. ‘I don’t know’, for example, makes correct use of the auxili¬
ary verb ‘do’ and has post-verbal negation. In this respect learners’ formulaic
speech contrasts markedly with their early creative utterances, which fre¬
quently show little evidence of any knowledge of inflectional morphology or
grammatical functors (for example, ‘No like it’ = ‘I don’t like it’). It is this that
enables researchers to distinguish'formulaic from creative speech in a corpus
of learner language.
The particular formulas learnt by individual learners are likely to vary, but
some seem to figure in just about every learner (for example, ‘I don’t know’
and ‘What’s this?’). Each formula is closely tied to the performance of a par¬
ticular language function which is communicatively important to the learner.
Hakuta (1986) notes that Uguisu’s formulas enabled her to express functions
which would have been beyond her if she had relied on her rule-governed
knowledge. In Ellis (1984b), I observed that the three classroom learners I
studied learnt formulas to enable them to meet their basic communicative
needs in an ESL classroom, where English functioned as the medium of in¬
struction. Krashen (1982) claims that formulaic speech occurs when learners
are forced to speak before they are ready and that, left to their own devices,
they will remain silent. But Krashen’s view seems to ignore the fact that it is
perfectly natural for any language user to seek to simplify the burden of pro¬
cessing language—by using formulas to establish ‘islands of reliability’, for
example. Learners, like native speakers, learn formulas because it reduces the
learning burden while maximizing communicative ability.
It is not surprising, therefore, to find that learners sometimes use the for¬
mulas they have acquired in unique ways. For example, they do not always
use them to perform the same functions as native speakers. Huebner (1980)
provides a detailed account of how an adult Vietnamese learner of L2 English
used two formulas ‘waduyu’ (= ‘what d’you’) and ‘isa’ (= ‘is a’). ‘Waduyu’
served initially as a general question marker. It was used ubiquitously where
native speakers would employ a variety of uT?-question words (‘what’, ‘how’,
‘why’, and ‘when’). Over time the learner gradually and systematically re¬
placed ‘waduyu’ with these target forms. ‘Isa’ functions initially as a topic
marker, but it was only used when the topic was not an agent or object and
was not identical to the topic of the immediately preceding sentence. The gen¬
eral point that Huebner makes is that the acquisition of the functions per¬
formed by formulas may be an evolutionary process.
A key question is that posed by Hakuta (1976): ‘To what extent do these
routines and patterns facilitate or hinder the acquisition of TL grammar?’ A
 Developmental patterns: order and sequence 87

number of researchers have suggested that formulaic speech serves as the ba¬
sis for subsequent creative speech when the learner comes to realize that utter¬
ances initially understood and used as wholes consist of discrete constituents
which can be combined with other constituents in a variety of rule-bound
ways. Clark (1974) illustrates how in first language acquisition new struc¬
tures can result from the juxtaposition of two routines or from the embedding
of one within another. Wong-Fillmore (1976), in a study of five Spanish¬
speaking learners of English, suggests that in L2 acquisition, formulas are
slowly analysed, releasing constituent elements for use in ‘slots’ other than
those they initially occupied. For example, to begin with, Nora (the fastest of
Wong-Fillmore’s learners) used two formulas:

I wanna play wi’ dese.

I don’ wanna do dese.

and then discovered that the constituents following ‘wanna’ were

interchangeable:

I don’ wanna play dese.

I wanna do dese.

She comments that this ‘formula-based analytical process ... was repeated in
case after case’ (1976: 645). In Ellis 1984b, I too found evidence to support
the view that formulas are worked on systematically by learners. I demon¬
strated how the ‘I don’t know’ formula was built on by combining it with
other formulas:

That one I don’t know.

I don’t know what’s this.

It was also broken down, so that ‘don’t’ came to be used in similar but differ¬
ent expressions:

I don’t understand.
I don’t like.

‘Know’ was eventually used without ‘don’t’:

I know this.

and with subjects other than ‘I’:

You don’t know where it is.

It is possible, therefore, that formulas are slowly unpackaged, releasing valu¬

able information, which is fed into the knowledge system the learner uses to
produce and understand creative speech.
A rather different view is taken by Krashen and Scarcella (1978). They ar¬
gue that formulaic speech and rule-created speech are unrelated. According
to this view, learners do not unpackage the linguistic information contained
88 The description of learner language

in formulas, but internalize L2 rules independently through attending to in¬

put. Their conclusion is that ‘the use of routines and patterns is certainly part
of language, but it is probably not a very large part’. This view is supported by
Bohn (1986), who analysed data that Wode (1981) collected from four Ger¬
man children learning English in the United States. This study found that
these child learners relied on formulaic speech very little in most situations,
although Bohn admits that they may not have been subject to the same com¬
municative pressures as other L2 learners. Formulaic-like speech was evident
when the children were involved in playing games, leading Bohn to suggest
that the prevalence of formulas in Wong-Fillmore’s subjects’ speech was an
artefact of the fact that she collected most of her spontaneous data during
games. Bohn concludes that formulas serve only ‘short-term production
tactics’ and play no role in acquisition.
It is not easy to choose between these two interpretations of the role played
by formulas. McFaughlin (1985), in a lengthy discussion of formulas, comes
down on the side of Wong-Fillmore, at least where children are concerned. In
considering this controversy, we should note that it is not easy to make a clear
distinction between formulaic and creative speech. In many instances, learn¬
ers seem to make use of patterns which are varied to a greater or lesser extent
through lexical substitutions. Such speech has both formulaic and creative
elements, suggesting that ‘we should move beyond misleading dichotomies
such as prefabricated formulas versus creative constructions’ (Rescorla and
Okuda 1987: 293). Even if Krashen and Scarcella are right and formulaic
frames do not evolve into productive rules, they almost certainly underestim¬
ate the importance of formulas for E2 learners. As Pawley and Syder have
shown, the task facing the learner is not just that of acquiring a rule system
but also of mastering a set of lexicalized sentence stems that will enable them
to process language efficiently. The development of target-like E2 ability,
then, requires the memorization of a large set of formulaic chunks and
patterns.

Structural and semantic simplification

In comparison with formulaic speech, the learner’s early creative utterances

are typically truncated, consisting of just one or two words, with both gram¬
matical functors and content words missing. Hanania and Gradman (1977)
give the following examples produced by their adult subject, Fatmah:
library (= He is in the library)
clean floor (= Give me something for cleaning floors)
come back (= He is coming back).

Pienemann (1980) gives similar examples from the speech of children learn¬
ing F2 German:

ein junge ball weg (= Ein Junge wirft den Ball weg)
 Developmental patterns: order and sequence 89

ein madchen bier (= Ein Madchen kauft Bier)

zwei kinder (= Es gibt zwei Kinder).

In Ellis 1984a, I found further evidence of simplified speech in the speech of

three children learning English in a classroom setting:

me no blue (= I don’t have a blue crayon)

eating at school (= She eats meat at school).

Such speech, therefore, is very common in the unplanned speech of both child
and adult learners.
These utterances, which bear a strong resemblance to those found in pidgin
languages, indicate that both structural and semantic simplification are tak¬
ing place. Structural simplification is evident in the omission of grammatical
functors such as auxiliary verbs, articles and bound morphemes like plural -s
and past tense -ed. Semantic simplification involves the omission of content
words—nouns, verbs, adjectives and adverbs—which would normally occur
in native-speaker speech (see Ellis 1982). Both structural and semantic simpli¬
fication may occur either because learners have not yet acquired the necessary
linguistic forms or because they are unable to access them in the production of
specific utterances. In other words, they may reflect processes of language
acquisition or of language production.6
Corder (1981b) argues that it may be misleading to talk of ‘simplification’
in such utterances:

... if in the process of first or second language acquisition the learner dem¬
onstrates that he is using a simple grammar or code, as is well attested, then
he has not arrived at that code or grammar by a process of simplification of
the target code. In other words, you cannot simplify what you do not pos¬
sess (1981b: 149).

According to this view, it is appropriate to refer to learner language descript¬

ively as ‘simple’, but is not appropriate to talk of ‘simplification’ by the learn¬
er. There are, however, a number of problems with Corder’s position. First,
when the learner produces an utterance like ‘hitting’ (= ‘He hit me’), it is pos¬
sible that the learner is simplifying for purposes of production (i.e. does have
the necessary knowledge to say ‘He hit me’ but does not do so because it is too
difficult in the particular communicative circumstances). Second, it is also
possible that an utterance like ‘hitting’ is indicative of an acquisitional pro¬
cess. The learner may have acquired this particular form because it is easy to
perceive in the input, in a sense, ‘simplifying’ the input by attending to and in¬
ternalizing a feature early on because it is more salient. It would seem appro¬
priate, therefore, to continue to talk of structural and semantic simplification.
Structural simplification can be described by means of the traditional cate¬
gories of a descriptive grammar. Semantic simplification is best accounted for
in terms of the descriptive categories provided by a case grammar (Fillmore
90 The description of learner language

1968; Greenfield and Dent 1980). For example, if a learner wishes to encode
the proposition:
He is hitting me
which involves these semantic categories:
(Agent) (Action process) (Patient)
the message can be conveyed by producing any one of these abridged
versions:
Hitting (= Action process)
He hitting (= Agent + Action process)
Hitting me (= Action process + Patient)
He me (= Agent + Patient).
Which version the learner chooses will reflect: (1) the linguistic resources
available or readily accessible, and (2) which constituents will be maximally
informative in context.
There is some evidence to suggest that learners, particularly children, tend
to begin speaking first in single-word utterances and then in increasingly lon¬
ger utterances, many of which are novel (see Saville-Troike 1988). However,
the nature of this progression is not as well-defined as in LI acquisition, per¬
haps because L2 learners have more developed processing capacities. Often,
utterances that manifest structural and semantic simplification are used along
with others that display little or even none.
One interesting possibility in the case of semantic simplification is that
learners follow some kind of order in acquiring the various semantic roles de¬
scribed by the case categories—as has been shown to occur in LI acquisition.
But to date there is little evidence to support such a claim, as few studies have
attempted to use case grammar to describe learner language (see Ellis 1982
and 1984a for one such attempt). The question of the order of acquisition of
structural features is considered in the sections that follow.
*

The acquisition of morphemes: order and sequence

As we have already noted, it is possible to consider acquisition in terms of
both the order in which different features are acquired and also the sequence
of stages evident in the acquisition of a single feature. Much of the early re¬
search focused on the order of acquisition. Subsequent research has increas¬
ingly paid attention to sequence.

The morpheme studies

In the 1970s a number of studies, commonly referred to as the morpheme stu¬

dies, were carried out to investigate the order of acquisition of grammatical
 Developmental patterns: order and sequence 91

functors such as articles and inflectional features such as plural -s. These stu¬
dies were motivated by similar studies in LI acquisition (see the previous sec¬
tion on LI acquisition). In particular, they sought to establish whether, as in
LI acquisition, there was an invariant order of acquisition. There were both
cross-sectional and longitudinal studies, although the former predominated.
The studies employed obligatory occasion analysis in order to establish the
accuracy with which learners of L2 English performed a range of morphemes.
In the case of cross-sectional studies, an accuracy order was calculated and
this was equated with acquisition order by some researchers, on the grounds
that the more accurately a morpheme was used, the earlier it must have been
acquired.
In two early studies, Dulay and Burt (1973; 1974c) investigated Spanish
and Chinese children, eliciting ‘natural’ spoken data by means of the Bilin¬
gual Syntax Measure (BSM—see Chapter 2, page 50 for a description of this
instrument). They found that the ‘acquisition order’ for a group of English
morphemes remained the same irrespective of the learners’ LI or of the meth¬
ods they used to score the accuracy of the morphemes. Bailey, Madden, and
Krashen (1974) replicated these studies with adult subjects, again using the
BSM. They found an ‘acquisition order’ that correlated significantly with
those obtained by Dulay and Burt. Larsen-Freeman (1976b) extended these
studies in two ways. First, she used learners with a wider range of Lis (Arabic,
Spanish, Japanese, and Farsi) and second she used five tasks to collect data:
the BSM, a picture-cued sentence repetition test, a listening comprehension
task, a multiple-choice reading cloze test, and a writing test involving filling in
blanks. She found that the learners’ Lis made little difference to the accuracy
orders she obtained. However, there were some differences in the orders for
the different tasks. The most notable was that between the speaking/imitation
tasks and the reading/writing tasks. Some morphemes (for example, plural -s
and third person -s) rose in the rank order in the reading and writing tasks.
One possibility, then, is that different orders exist for oral and written learner
language. This finding is not problematic where production is concerned, as
speaking and writing are influenced by different sociolinguistic and psycho-
linguistic conditions. It becomes problematic if the goal is to determine a sin¬
gle, invariant order of acquisition that is distinct from actual use. However,
Krashen, Butler, Birnbaum, and Robertson (1978) found that the accuracy
orders obtained from written data did correlate significantly with those re¬
ported by Dulay and Burt for oral data. In this study, two kinds of writing
task were used—one requiring ‘fast’ writing and the other ‘careful’ writ¬
ing—but neither encouraged attention to discrete items as was the case with
Larsen-Freeman’s writing task. The distinction between ‘free’ and ‘careful’
writing did not affect the morpheme orders. These studies are summarized in
Table 3.4. They constitute only a small selection of the total morpheme
studies that were undertaken.
92 The description of learner language

o C

follows observed order of

■ft C\J 0 o
0

instruction is that which

0 —J c
-Q c
'(0 N ® p 0
0 3 0 •p 03 C O-Q
Is r 0 2Z
0 > ■»-

3 The most effective

.■t; o O .O C
§ 1 O 0 0 ^ *5 -g ;+= 0 F 0
.ts I S'
© c 0
□ fc.s
^ 0
O) „ 0 5 F
E
0
5
Q-
0 0 ^
i §3
0) ^
-Q <d O’ 0 <f 0 o
0 p t O 0 0 5
T3 0
O
O g
-O
rj
C 0 «
0 O 0 03 o Z 0 E if E
o 0 0 0 0 El
C

difficulty,
Z
o E|
t° E ov
C 0 c b: C CL 5 w c o raE> 3 0 C\J Q. ® 1
m 03—1
'55 Q) 2 ®r'w g ,® CD 0 0 0
®Z
D
O 0) 5 s |a E it iS 0 0 > 5T ^ ® 1- 3. - o 3 •5 «
O 0 o 0 TJ ±3 r- "O to Z
c f— C o E UJ U (flv h- 0 T3 w ? E £ < 0 —J <■0 0
o
O C\J CM

CO 0
■o

those reported for all but

. "O
® ® >. o

one of Dulay and Burt’s

0 0

different from LI order.

B) £
0

3 The adult orders were

0 0 I* 2D 2Z
0 0 ® o
■D ^ D) ■o TO 0
0 — ■8 B1
2 ^ c
£ ■° O to
0 o
0 i_r t~ 0
O 2Z 0 0
2Z 0 0 03 I E O -D ?!
Si c
Q. 0 -J >
c 0 > c £ m
o o Q.

(1973) groups.
0 . 3 .2 o >» c r) if
O 0
E
»
'5
S I! «
5) 2 £ :s a 0
o
0
0
°
O
0
m
0
0
w ® O o 0
0
a O’ <i> >*4- E
rj)
'0
0
E
0 ( Q_-C
o o o P
C 0
C 0 as
0 03 -Q 2Z
c
0 ®
+- 0i«0-2
< 0 0
^ TJ >
« -C i- Q_ 0 o. O -
sO 0
0s > .x 0 hr 0^ 0 •_ 0
ID 0 ® 0 0
D 2Z O 0 O . . O CL IB?
n CD
o0 00 -o £ -C "0 "o E I- -Q £ E £ 0 h- 0

cc

0 0
3 3
O) 03 03
_c _C

c m 0 CD 0 CD 0
o
o 12 3S I 3
© 9= 0 £ $
o
0 ^
■4-* 52
o 0 X
0
5 3 TD 0
_ -t—•
-§ 3
_ -*-•
ft sf 2 £ e 5.
Q O co O CD O CO

c c
0 0
>_ u
>% 0
— ? 0 *o 0
00 z z
o o
L_ —
»_ • -
> 0 03 0
0
2 g> 03 03 a- -E —
°l . .g .£ $ 0^0
0
"m
0 *rrt ^
0 rr, CD
(DO)0?
Q.
2 0* -=
0
Cl Cl (L
0 0
■O
0
fn ' 00
r-
I
0 «2 0
» o
0 0 2
0 2 E' 0
03
0 TJ *“ 0
C ±2 0 Q. 0
o 3 H=
0 <3 S. ® _03
15* S-Wp CD O £ ■O
0
^
0 c t:
o?
■§
3 o in o co i5
(0 co in .Q o

c
0
2Z
0
0
i—
2D
CO "vf T3
N C
03 03 0
T“
C
t 0
3 ■o
CD CO ■o
"O ■O 0
C c
0 0
>
■o 0 _0 0
3
3 3
35 D Q m t-
 Developmental patterns: order and sequence 93

S'.52 ■§£
O)
c
c -C
0 O « o
p o
"D —
=3
cr 0
0
o
o c
S> :8 ®
0 CO 0 Q-v
*o o 05 .v- 0
3 "O
I ®i
g? 5
-g.c0 « = 5
CO Jr 0-0
0C E0
®CO 0b ■§CO£ r- O 0 ©S 0 £P
- -P 0
E'B 8 o 00-PO £
£E 0o 0o .9CO CL
+J
0
c CO E
0 2>*- o^ c0 «5 o0 •= £0 0
o CO 0 0 5=
C
0 O
c A
-c S
■</>
O
-C 00 "O 0 E cr co
J3 ® a Q. 1 C 13 J2 0
CL o ^ 03
o o © >-O 0 oo Eo wp
c
o ll E
Q-l— 5 o o E 0 « s £5il
o OsJ C\J

c
CO c
o E ® I sz £
O0 gs
v CO O
■o0
h= 0 o E -Q
c © S-D 0 E 0 0 •c 0
.9> 0 <0 "q.^
® B- £ : 1 "E
O bo If 0 03 Q- "c/5 <3
0
^£
03a) -oco p0 o ° 0 _p CO CO
0DM »
0 0^ oo
1 t; ® M
E "o
> >> r . ® o\
0 0 0 C o *T H—
0 ' u "O C O Sm-
0
»m b 0 c sz
<s Mb % ■5 -*-*
I
© 0 JZ
"TZ

J C O -Q3 ‘-PO !i: ?®c >n o-B

CO 0 o
0 O C C
3-E 0c 0<5 0
o" 0 0
„ •
(- v
—
tq
OT 2 <2 © *© 0
13
0 o £ -a >,
jLJ
T=- 0 ® CO0 -gO ;I >-« o
^

cE -S CO £ ■! Q=3 <O ^ CL I—0? 4=0f~ 0t: 0 •“ P 13 “ — T3 T3

0 =3
0 I— > 0 0 ■
3
CO LU Q o
^ .E -o .
o
0
CC

"Q
■O
c a
O) :* c
;; IB
i .£c ccn — o
■ 0 p
s
:; JJ« *” -a
>.i b
j o o
■ c 0
£

i-i?
3 g 0 O C\J 0

j . d) is o
H—
^ D) .£
C _* o ~ 0O
o E

- T3 CO 0 0
CO eI
j; ir 9-
3 0
© n
o 0 -- o

5
V)
0

- 'c0
0

.9 a life
-Q GO o C 0
0
< 0 0
II 0 §>1
0 0 0 0• ^ CO
*o3 -C>
5.°- c c 11' 0
<n ®"
£ co
5 a 0
03 0 C
o
0
3 ® 0
3
i— i—
0 O
0 o>>C ^0
3 ^ 0 I!0^
o C
S ©3O
C/> C\i ”3

E
3
n 0
-Q
CD
h-
03
T-
.i?
CD 03
c
0
E Is
D 0
0
0 CD tr
. 0
c -Q
0 O

*o>
-c CC
co
3 co ^
55 i 0
94 The description of learner language

One of the problems of rank orders is that they disguise the difference in ac¬
curacy between various morphemes. Thus a morpheme that is just 1 per cent
lower than another morpheme is given a different ranking in just the same
way as a morpheme that is 30 per cent lower. To overcome this problem,
Dulay and Burt (1975) and later Krashen (1977) proposed a grouping of
morphemes. They argued that each group constituted a clear developmental
stage in that the morphemes within it were ‘acquired’ at more or less the same
time. Figure 3.1 presents Krashen’s ‘natural order’ of morpheme acquisition.
The picture that emerges from these studies is of a standard ‘acquisition or¬
der’ that is not rigidly invariant but is remarkably similar irrespective of the
learners’ language backgrounds, of their age, and of whether the medium is
speech or writing. A different order occurs only when the learners are able to
focus on the form rather than the meaning of their utterances. As Krashen
(1977: 148) put it, whenever the data reflects a focus on meaning there is ‘an
amazing amount of uniformity across all studies’.
A number of later cross-sectional morpheme studies have been carried out,
including some of languages other than English (for example, Van Naerssen’s
(1980) study of L2 Spanish). One of the best is Pica’s (1983) study of the
morpheme orders of three separate groups of L2 learners—a ‘naturalistic’
group, an instructed group, and a mixed group. We will consider this study in
some detail in Chapter 14. Flere we simply note that Pica—who was careful
to claim only an ‘accuracy order’—found the same ‘natural order’ even when
she took account of learners’ over-use of morphemes in her scoring procedure
(i.e. used target-like use analysis rather than obligatory occasion analysis).
This is important because it helps counter one of the main criticisms levelled

Figure 3.1: Proposed ‘natural order’ for L2 acquisition (Krashen 1977)

 Developmental patterns: order and sequence 95

at the morpheme studies, namely that they have failed to consider inappropri¬
ate morpheme use in non-obligatory contexts.
There have been fewer longitudinal morpheme studies. Rosansky (1976)
examined the order of acquisition of the same morphemes investigated by
Dulay and Burt in the speech of Jorge, a Spanish learner of English, over a
period of 10 months. She concluded that the order of acquisition of this indi¬
vidual learner did not conform to the ‘natural order’ reported for groups of
learners. Also, when she carried out cross-sectional analyses with data col¬
lected at different points in time she found varying orders. Hakuta (1974) col¬
lected spoken data from a five-year-old Japanese girl, Uguisu, over several
months. The acquisition order he obtained did not match that of Dulay and
Burt. Articles, for instance, which have a high ranking in the cross-sectional
accuracy orders, had a low ranking in the acquisition order for Uguisu.
Schmidt’s (1983) study of Wes, a Japanese painter living in Hawaii, also
found some discrepancies with the cross-sectional order (plural, article, and
past regular having lower ranks). One possible explanation for the difference
in results between the cross-sectional and longitudinal studies is that the res¬
ults of the latter were misleading because accuracy levels were sometimes cal¬
culated using fewer than ten obligatory occasions, a point that Krashen
(1977) emphasizes. Nevertheless, the differences do cast some doubt on the
validity of equating cross-sectional accuracy with acquisition orders.
The morpheme studies have been subject to some stringent criticisms in ad¬
dition to the doubts about using accuracy order as a basis for discussing ac¬
quisition (see Hatch 1978d and 1983a; Long and Sato 1984). One criticism is
that the method of scoring morphemes does not take account of misuse in in¬
appropriate contexts. But, as we have seen, Pica’s (1983) study suggests that
even when overuse is taken into account it does not affect the order. Another
criticism is that the use of rank order statistics hides meaningful differences,
but Krashen’s grouping of features into a ‘natural order’ (see Figure 3.1) goes
some way to overcoming that objection. Other objections are that the re¬
search has been restricted to a small set of morphemes, that the morphemes
studied constitute a rag-bag of disparate features (the acquisition of ‘articles’
and ‘3rd person -s’, for instance, pose the learner very different tasks, as one
involves semantic considerations and the other is a purely formal feature),
and that the research has lacked theoretical motivation. These criticisms are
generally acknowledged, with the result that morpheme studies have now
been discontinued. However, Larsen-Freeman and Long (1991) conclude
their own survey of these studies with the claim that they provide strong evid¬
ence of a developmental order:
Contrary to what some critics have alleged, there is in our view too many
studies conducted with sufficient methodological rigor and showing suffi¬
ciently consistent general findings for the commonalities to be ignored
(1991:92).
96 The description of learner language

Such a conclusion, however, appears overly charitable, as it fails to recognize

the most serious limitation in the morpheme studies—the conceptualization
of acquisition in terms of what Rutherford (1988) calls ‘accumulated enti¬
ties’, i.e. the mastery of grammatical items one at a time. The inadequacy of
this view will become clear when we take a closer look at how learners ac¬
quire individual morphemes.

The acquisition of individual morphemes: pronouns

A close look at individual morphemes shows that they are acquired gradually
and systematically. Learners do not progress from a state of non-acquisition
to a state of acquisition, but rather pass through a series of stages. In Chapter
4 we shall discuss the extensive research that has looked at individual morph¬
emes; here, we will consider just one grammatical feature—pronouns
—drawing on research which has investigated the acquisition of this feature
in several languages.7
In the pronoun system of a language, a number of semantic distinctions are
to be found. For example, English distinguishes gender (for example, ‘he’ and
‘she’), person (for example, ‘I’ and ‘you’), number (for example, ‘he’ and
‘they’), and case (for example, ‘I’ and ‘me’). A further distinction between per¬
sonal pronouns and possessives (for example, ‘him/her’ v. ‘his/her’) can also
be made. Learners also have to discover when pronouns can be omitted and
when it is obligatory to use them. Not surprisingly, pronouns present the
learner with a challenge that can only be met over time. Felix and Hahn
(1985) note that nearly all studies of naturalistic acquisition provide evidence
of frequent omission of pronouns and overgeneralization of individual pro¬
nouns during the early stages. This is evident in all the languages which have
been studied.
In the case of English—an example of a language where pronouns are ob¬
ligatory in many contexts—learners are initially still likely to produce utter¬
ances with no pronouns:

Kicking (= He kicked me)

Felix and Hahn (1985) suggest that the various semantic features are then sys¬
tematically acquired and propose a tentative ordering. Learners begin by dis¬
tinguishing just one feature, usually person. During this stage, they may
distinguish only first person from all other persons. Thus ‘I’ or ‘me’, or some¬
times both interchangeably, are used to realize first person, while a single pro¬
noun (such as ‘you’ or ‘he’) refers to all other persons. Later, they distinguish
pronouns according to number. Third person pronouns are added next, but
no gender distinction is made. This is acquired last. A similar sequence of ac¬
quisition is evident for possessives by French learners of English (see Table
3.5). Butterworth and Hatch (1978) illustrate the kind of substitutions and
overgeneralizations which result from this process of acquisition in their
 Developmental patterns: order and sequence 97

Stage Description Examples

1 The use of definite articles in place of ’She reads the book.
possessives.
2 The use of a generalized possessive for all ’She reads your book.
persons, genders, and numbers.
3 The use of a single third person possessive, ’She reads his book.
which is overgeneralized.
4 Differentiated use of possessives with some ’She reads her book to her
possessed nouns but continued errors with brother.
human nouns.
5 Correct use of possessives with all nouns. She reads her book to her
brother.

Table 3.5: The acquisition of possessive pronouns by francophone learners of

English (table based on information provided in Lightbown and Spada 1990)

subject, Ricardo, an adolescent learner of English. When they first started stu¬
dying him he used both ‘me’ and T for subject pronoun. He failed to distin¬
guish personal and possessive pronouns, producing errors like:

* You girlfriend (= Your girlfriend)

*He car (= His car).

Ricardo used ‘he’ and ‘she’ infrequently, often substituting them with a noun,
perhaps because he found it easier to simply repeat a noun than to substitute it
with a pronoun. ‘We’, ‘it’, and ‘they’ were also omitted or replaced with ‘he’.
Further evidence of developmental regularities in pronoun acquisition
comes from a study by Gundel and Tarone (1983). This investigated five
adult learners of L2 English (three with Spanish as their LI and two with Chi¬
nese) and utilized data from a variety of sources (tape-recorded
conversations, a recorded picture description task, and two kinds of gram¬
matically judgement tasks). It also reports an analysis of data collected from
grade 1 learners of L2 French in an immersion program in Canada. The main
focus of the study was the learners’ omission of object pronouns in obligatory
contexts: for example, ‘" He didn’t take’.
In the case of the Chinese learners, the lack of use of object pronouns is at¬
tributable to transfer, as object pronouns are not obligatory in Chinese. How¬
ever, the Spanish learners of English and the English learners of French made
more such errors than did the Chinese ones, despite the fact that object pro¬
nouns are obligatory in their Lis. Gundel and Tarone suggest that these
learners begin with a transfer hypothesis (i.e. that the TL will have object pro¬
nouns in the same position as the LI), but reject this when they find that
the L2 input is not consistent with it. This leads them to construct a second
hypothesis, that the L2 has no object pronoun forms at all. As the L2 input
also belies this hypothesis, they abandon it and form a third hypothesis: that
98 The description of learner language

the L2 does have object pronouns but not in the same positions as in their Lis.
Once again, then, we see evidence of the systematic acquisition of a particular
feature—in this case the use of object pronouns.
Similar patterns of pronoun acquisition are evident in other languages.
Klein and Rieck (1982, cited in Klein 1986) used data from a sentence repeti¬
tion task administered to Spanish and Italian migrant workers to investigate
the acquisition of pronouns in L2 German, where, as in English, they are ob¬
ligatory in many contexts. They note that the first/second person singular dis¬
tinction is acquired early and the number distinction later. Nominative case
forms (for example, ‘ich’) are acquired before accusative and dative case
forms (for example, ‘mich’ and ‘mir’). For example, when asked to repeat the
sentence ‘Vielleicht hat sie ihn zu Hause bei ihren Eltern vergessen’ (= ‘Per¬
haps has she it at home with her parents forgotten’) the least advanced learn¬
ers omitted both ‘sie’ and ‘ihn’ while the most advanced learners included ‘sie’
(= subject) but omitted ‘ihn’ (= object). This occurred despite the fact that the
learners’ Lis permit deletion of subject pronoun but require retention of ob¬
ject pronoun. Klein also notes the tendency for the learners to replace pro¬
nouns with nouns.
The acquisition of pronouns in Dutch has been investigated by Broeder,
Extra, and van Hout (1989) in a study of four adults that was part of a large
international project directed at investigating untutored L2 acquisition by ad¬
ult immigrants in a number of European countries (the European Science
Foundation Second Language Project on Adult Second Language Acquisi¬
tion). The main findings were that subject pronoun forms were acquired first,
followed by object forms, and then possessive forms. Also, singular pronoun
forms were acquired before plural forms. This study found no clear evidence
of LI influence, leading the researchers to conclude that ‘target language re¬
lated principles were ... more influential determiners of acquisition than
source language related principles’ (1989: 104).
Finally, we consider pronoun acquisition in Spanish. Andersen (1983a)
provides a detailed discussion of the acquisition of clitic pronouns (i.e. pro¬
nouns like ‘lo’, ‘me’, and ‘se’, which function as direct objects, indirect ob¬
jects, or as reflexives) by one native speaker of English, Anthony. The
following are some of his findings. Anthony tended to use nouns in place of
clitic pronouns. Lie showed greater control of first person ‘me’ and ‘nos’ than
of third person singular pronominal forms (‘lo’ and ‘la’). He omitted plural
third person forms (‘los’ and ‘las’). He was able to position first person ‘me’
correctly before the verb but experienced difficulty with the third person pro¬
nouns in this position. He had no difficulty in positioning pronouns after the
verb.
The commonalities in these accounts of pronoun acquisition are striking.
They suggest that learners of different languages (English, French, German,
Dutch, and Spanish) experience similar problems with pronouns and solve
them in similar ways. In many studies, there is evidence that the learner’s LI is
 Developmental patterns: order and sequence 99

a factor in Anthony’s preference for post- rather than pre-verbal pronoun

positioning, for instance. But there is also strong evidence of ‘universal’ pat¬
terns of acquisition, at least where these five Indo-European languages are
concerned. To substantiate claims regarding universality, of course, it will be
necessary to obtain evidence from a wider range of languages.

The acquisition of syntactic structures

We will look at three sets of syntactic structures to see to what extent they
provide support for the existence of developmental patterns in second lan¬
guage acquisition. First, we will review a number of early studies which ex¬
amined the acquisition of negatives in English and German and which
provide evidence of a clear sequence of development. Second, we will con¬
sider relative clauses, which provide evidence of an order of acquisition.
Third, we will look at the acquisition of word order rules in L2 German. This
provides evidence of both an acquisition order (as different TL rules are ac¬
quired one after another) and also of a developmental sequence (as learners
also manifest transitional structures which differ from the TL norms).
Finally, we will look at an attempt to characterize general grammatical devel¬
opment in an L2.

The acquisition of negatives in English and German

The development of negatives in English outlined below is based on several
early studies (Ravem 1968; Milon 1974; Cazden et al. 1975; Wode 1976 and
1981; Adams 1978; Butterworth and Hatch 1978). These studies cover learn¬
ers of English with Japanese, Spanish, German, and Norwegian as their LI,
and also children, adolescents, and adults. Schumann (1979) provides an ex¬
cellent overview of a number of studies involving Spanish learners of English.
Negation is an example of a transitional structure (see Dulay, Burt, and
Krashen 1982). That is, it involves a series of forms or structures which learn¬
ers use en route to mastering the TL form. These interim forms are indicative
of the developmental stages that learners pass through on the way to TL com¬
petence. Initially, negative utterances are characterized by external negation.
That is, the negative particle (usually ‘no’ but sometimes ‘not’) is attached to a
declarative nucleus:

No very good.
No you playing here.

A little later, internal negation develops; that is, the negative particle is moved
inside the utterance. This often coincides with the use of ‘not’ and/or ‘don’t’,
which is used variably with ‘no’ as the negative particle. ‘Don’t’ at this stage is
formulaic as it has not been analysed into its separate components ‘do’ and
‘not’.
100 The description of learner language

Mariana not coming today.

I no can swim.
I don’t see nothing mop.

A third step involves negative attachment to modal verbs, although this may
again occur in unanalysed units initially:

I can’t play this one.

I won’t go.

In the final stage, the TL rule is acquired. The learner has developed an auxili¬
ary system and uses ‘not’ regularly as the negative particle (i.e. the use of ‘no’
with verb is eliminated). Negative utterances are now marked for tense and
number, although not always correctly:

He doesn’t know anything.

He didn’t said it.
She didn’t believe me.

This sequence is summarized in Table 3.6.

Stage Description Example

1 External negation (i.e. ‘no’ or ‘not’ is placed at the No you are playing here.
beginning of the utterance).
2 Internal negation (i.e. the negator - ‘no’, ‘not’ or Mariana not coming today.
‘don’t’ is placed between the subject and the
main verb).
3 Negative attachment to modal verbs. 1 can’t play that one.
4 Negative attachment to auxiliary verb as in target She didn’t believe me.
language rule. He didn’t said it.

Table 3.6: Summary of general stages in the sequence of acquisition in L2 English

negation

The way along this route is a gradual one. Some learners can take longer
than two years and some never travel the whole distance. The stages are not
clearly defined but overlap considerably. Development does not consist of
sudden jumps, but of the gradual reordering of early rules in favour of later
ones. There are also some differences among learners, reflecting their LI.
These are discussed in Chapter 8. Individual learners display different prefer¬
ences in the choice of negative particle—‘no’ is the most common, but ‘not’
and unanalysed ‘don’t’ have also been attested.
German negation differs from English in a number of respects. Like Eng¬
lish, German places the negative particle after the finite verb:

Ich kann nicht kochen (= I can not cook).

Ich habe nicht gekocht (= I have not cooked).
 Developmental patterns: order and sequence 101

but unlike English, German does not have a dummy auxiliary, so ‘nicht’ fol¬
lows the main verb in sentences like:

Ich koche nicht (= I cook not).

Also, when there is a direct object and a ‘free’ adverb, ‘nicht’ follows them as
in:

Ich habe den Wagen gestern nicht gefahren.

(= I have the car yesterday not driven).

Studies of the naturalistic acquisition of sentential negation in L2 German

include those by Felix (1978), Lange (1979), Pienemann (1981), and Clahsen
(1982). Weinert (1987) and Eubank (1990) provide summaries of the general
pattern of development. The first stage involves sentence external negation,
using ‘nein’ usually in sentence-initial position:

nein helfen (= no help)

nein spielen Katze (= no play cat).

Learners then move on to internal negation, placing the negator in both pre¬
verbal or postverbal position. Main verbs tend to be negated preverbally and
other verb types postverbally:

ich das nicht mach (= I that not do)

ich kann das nicht (= I can that not).

As learners master internal negation, they eliminate preverbal negation, posi¬

tioning the negative particle after the verb even with main verbs:

ich fallst nicht runter (= I fall not down)

In the final stage, learners master the position of negator after direct objects
and free adverbials (i.e. they are able to separate ‘nicht’ from the verb). Thus,
despite the differences in the ‘final states’ towards which learners of English
and German are targeted, marked similarities in the sequence of acquisition
of negatives in the two languages can be seen.

The acquisition of relative clauses in English and Swedish

The acquisition of relative clauses presents learners with two tasks. In the case
of English, they must first learn that relative clauses can modify noun phrases
that occur both before the verb (i.e. as subject of the main clause) and after the
verb (i.e. as object or in a prepositional phrase):

The man who lives next door is getting married.

We met the man who lives next door.

Second, they must learn the various functions that the relative pronoun can
serve. English permits a range of functions, as shown in Table 3.7. These two
tasks amount to a substantial learning burden. How do learners tackle it?
102 The description of learner language

Function Example

Subject The man who lives next door...

Direct object The man whom 1 saw ...
Indirect object The man to whom 1 gave a present...
Oblique (object of preposition) The man about whom we spoke...
Genitive The man whose wife had an accident...
Object of comparative The man that 1 am richer than ...

Table 3.7: Relative pronoun functions in English

Studies which have investigated the acquisition of relative clauses in L2 Eng¬

lish include those by Cook (1973), Schachter (1974), Ioup (1977), Gass
(1979), Schumann (1980), Chiang (1980), Gass and Ard (1980), Hyltenstam
(1984), Pavesi (1986), and Hansen-Strain and Strain (1989). The learners in
these studies again come from a variety of language backgrounds.
With regard to the first task, Schumann found that the five Spanish¬
speaking learners he investigated began by attaching a relative clause to a
noun phrase that follows the verb. Often learners include a pronominal
copy:8
I know the man who he coming.

in which case the relative pronoun may be functioning in a similar way to the
co-ordinator ‘and’, joining two main clauses. Only when learners omit the
pronominal copy can they be said to have acquired the use of relative clauses:
Joshua’s a boy who is silly.

Relative clauses modifying the subject of the main clause appear later:
The boys who doesn’t have anybody to live they take care of the dogs.
Gass^and Ard (1980) provide evidence for a different sequence of acquisition.
Where the data Schumann used consisted of naturally occurring speech, their
data came from a sentence joining task, which may explain the differences.
With regard to the second task, learners may begin by omitting the relative
pronoun (see Schumann 1980):
I got a friend speaks Spanish.
Next, they may use an ordinary personal pronoun:
I got a friend he speaks Spanish.
and finally the relative pronoun proper is used:
I got a friend who speaks Spanish.

The first function to be mastered is that of subject. The order of acquisition

then proceeds as shown in Table 3.7. Two studies, one of L2 Swedish
 Developmental patterns: order and sequence 103

(Hylstenstam 1984) and the other of L2 English (Pavesi 1986), both of which
used implicational scaling, could find no clear evidence for differentiating the
order of acquisition of indirect object/oblique and genitive/object of compari¬
son, however.
The retention of pronominal copies is also linked to the acquisition of the
functions of the relative pronouns. Hyltenstam’s study indicates that the ex¬
tent to which learners retain copies is influenced by their LI. Thus, Persian
learners, whose LI permits copies, are much more likely to retain them than
Linnish learners, whose LI does not. However, all learners use at least some
copies and, interestingly, are more likely to use them with the relative pro¬
noun functions that are difficult to acquire. Hylstenstam and Pavesi provide
evidence to suggest that copies disappear from learners’ language in the same
order as the relative pronoun functions are acquired.
The study of the acquisition of relative clauses is an important area of work
in SLA research because it has been used to test predictions based on marked¬
ness. We will return to it in Chapters 8 and 10. Also, we will examine several
studies which have investigated the effects of instruction on the acquisition of
relative clauses in Chapter 14.

The acquisition of German word order rules

A large number of studies have investigated the acquisition of German word
order rules in naturalistic L2 acquisition. One of the most important was the
Zweitspracherwerb Italienischer und Spanischer Arbeiter (ZISA) project,
which focused on learners with a romance language background. This has
been reported in a series of articles by Meisel, Clahsen, and Pienemann (for
example, Meisel, Clahsen, and Pienemann 1981; Clahsen 1980; Clahsen
1984; Clahsen, Meisel, and Pienemann 1983; Pienemann 1980). Subsequent
work has also examined Turkish learners (for example, Clahsen and
Muysken 1986). Jordens (1988) provides a helpful survey of the relevant
research.
A clear developmental pattern of the acquisition of German word order
rules emerges from this research. This is summarized in Table 3.8.
Pienemann, Johnston, and Brindley (1988: 222) claim that this pattern ‘is
now probably one of the most robust empirical findings in SLA research, be¬
cause the same sequence has been found with a considerable number of fur¬
ther informants in studies carried out independently of each other.’ The
pattern does allow for some variation, however. Thus, stage 1 (canonical or¬
der) may not necessarily consist of subject-verb-object (SVO) for all learners.
Whereas learners with romance language backgrounds begin with SVO,
other learners (for example, Turkish) begin with different basic word orders
that reflect their LI. Also, learners vary in how they progress through the se¬
quence. Some learners appear to move slowly, consolidating each new stage
before they move on to the next, whereas others move on to a new stage very
104 The description of learner language

Stage Name Description Examples

1 Canonical order Romance learners begin with die kinder spielen mim ball (=
SVO as their initial hypothesis the children play with the ball).
about German word order.
Adverbials appear in
sentence-final position.
2 Adverb preposing Learners are now able to da kinder spielen (= there
move an adverbial into children play).
sentence-initial position.
However, they do not yet
invert the subject and verb as
is required when a sentence
begins with an adverbial. This
is not acquired until stage 4.
3 Verb separation Learners move non-finite alle kinder muss die pause
verbal elements into clause- machen (= all the children
final position, as required in must the pause make).
the target language.
4 Inversion Learners learn that in certain dann hat sie wieder die knocht
contexts such as sentence- gebringt (= then has she again
initial adverbials and the bone bringed).
interrogatives, the verb must
precede the subject.
5 verb-end Learners learn that the finite er sagte dass er nach hause
verb in subordinate clauses kommt (= he said that he to
goes in clause-final position. home comes).

Table 3.8: Sequence of acquisition of German word order rules (based on

Pienemann, Johnston, and Brindley 1988)

rapidly, not bothering if they have not achieved a high level of accuracy in the
structure belonging to the prior stage. For example, in stage 4 (inversion)
some learners acquire the rule in one linguistic context (for example, with
sentence initial adverbials) and then move on to stage 5 (verb-end) before ac¬
quiring inversion in other linguistic contexts (such as interrogatives). Other
learners take time to acquire the rule in all its contexts before they progress to
verb-end. Differences among learners also exist with regard to their ultimate
level of achievement. Many of the learners in the ZISA project failed to reach
stages 4 and 5.

A general pattern of L2 grammatical development

In subsequent work, an attempt has been made to explain the developmental
pattern in the acquisition of German word order rules by means of a set of
cognitive processing operations that underlie the production of sentences
manifesting each word order rule. These will be described in Chapter 9 when
 Developmental patterns: order and sequence 105

Stage Main features Example

1 Single words; formulas My name is_.

2 SVO; plural marking 1 eat rice.
3 ‘Do’-fronting; adverb preposing; neg. Do you understand me?
+V Yesterday 1 go to school.
She no coming today.
4 Pseudo-inversion; yes/no inversion; V Where is my purse?
+ to + V Have you car?
1 want to go.
5 3rd pers. -s; do-2nd He works in a factory.
He did not understand.
6 Question-tag; adverb-VP He’s Polish, isn’t he?
1 can always go.

Table 3.9: Generalized pattern of acquisition for L2 English (based on

information provided in Johnston and Pienemann 1986)

we examine the Multidimensional Model. They are mentioned here because

the operations which have been identified provide a basis for predicting the
development of word order rules (and indeed other kinds of rules) in lan¬
guages other than German. Johnston and Pienemann (1986), for instance,
tested the predictions afforded by the model on the acquisition of English by
migrant workers in Australia and found evidence to support it. They used im-
plicational scaling on data collected cross-sectionally to establish the hier¬
archy of stages shown in Table 3.9. This constitutes one of the most successful
attempts to date to identify a general pattern of development.
Finally, the work on L2 German and English suggests that not all features
are developmental in the sense that their acquisition occurs at a particular
stage of learners’ overall development. Some are variational, i.e. they may or
may not be acquired by individual learners and can be acquired at any stage of
development. Meisel, Clahsen, and Pienemann (1981) give copula as an ex¬
ample of a variational feature, pointing out that learners vary enormously
regarding both whether and when they acquire it.9

The L2 = LI hypothesis
We have now reviewed some of the major features of LI and L2 acquisition
and so can turn to consider the L1=L2 hypothesis (also referred to as the iden¬
tity hypothesis). This has received considerable attention in SLA research as it
raises a number of important theoretical issues (see Clahsen 1990). These
concern whether the fundamental principles that underlie LI and L2 are the
same, and whether the language acquisition device which mentalists claim is
responsible for LI acquisition is available to L2 learners. A number of studies
have sought to compare learner language resulting from the two types of
106 The description of learner language

acquisition (for example, McNamara 1973; Dulay and Burt 1974c; Ervin-
Tripp 1974; Cook 1977; McLaughlin 1978a and 1985; Felix 1978; Ellis
1985b; Bley-Vroman 1988).
The similarities in learner language in LI and L2 acquisition are perhaps
most pronounced in the early stages of development. There is evidence of a
silent period, of the use of formulas, and of structural and semantic simpli¬
fication in both types of acquisition. However, there are also obvious differ¬
ences. Whereas all LI learners necessarily pass through a silent period, many
L2 learners—especially adults—do not. Many L2 learners appear to make
greater use of formulas than LI learners. Also, L2 learners are able to produce
longer and less propositionally reduced utterances from the beginning. A cor¬
rect characterization of early LI and L2 acquisition might be to say that L2
learner language displays many of the features of LI learner language plus
some additional ones. Felix (1978) points to a number of differences in a com¬
parison of sentence types in the acquisition of German. He found that the L2
children he studied produced only three different multi-word utterance types,
and argues that this contrasts with LI acquisition, where learners have been
shown to produce a multitude of different structures from the two-word
stage.
The morpheme order acquisition is not the same in the two types of acquisi¬
tion. Dulay and Burt (1974c) compared the ‘acquisition order’ they obtained
for nine English morphemes with the acquisition order for the same morph¬
emes obtained in both longitudinal studies (for example, Brown 1973) and
cross-sectional studies (for example, de Villiers and de Villiers 1973) of LI
English. They found that the orders were different. Articles, copula, and aux¬
iliary ‘be’ were acquired earlier by L2 learners, while irregular past tense was
acquired later. However, the L2 order they obtained did correlate with the LI
order obtained by Porter (1977), who used the same data collection instru¬
ment—the Bilingual Syntax Measure (BSM). This has led some researchers
(for example, Rosansky 1976; Hakuta and Cancino 1977) to suggest that
Dulay and Burt’s acquisition order is an artefact of the BSM.
The process by which individual morphemes are acquired displays both
similarities and differences. For example, both LI and L2 learners omit pro¬
nouns and they both overgeneralize individual pronouns. The substitution of
nouns for pronouns also occurs in both types of acquisition. However, LI
learners commonly substitute their own name in place of the first person sin¬
gular pronoun for example, ‘Lwindi eating’ ( = I am eating), which has not
been attested in L2 acquisition, except by very young children.
The similarities between LI and L2 acquisition are strongest in syntactical
structures. The evidence from studies of negation (and also interrogatives)
suggests that learners pass through a remarkably similar sequence of acquisi¬
tion for these structures. Readers are invited to compare the order of LI
acquisition for English negatives shown in Tables 3.2 and 3.6. The sim¬
ilarities are truly striking. However, once again, the sequences in the two
 Developmental patterns: order and sequence 107

types of acquisition are not identical. For example, children acquiring Ger¬
man as an LI begin with the verb in final position (Clahsen 1988):
wurs hier schnitt (= sausages here cut)

whereas, as we have seen, L2 learners begin with a canonical order derived

from their LI, which in the case of romance language learners results in an
SVO word order.
Given the mixed results available from comparative studies of LI and L2
acquisition, it is perhaps not surprising to find widely diverging conclusions
regarding the L2=L1 hypothesis. Bley-Vroman (1988) emphasizes the differ¬
ences between LI and adult foreign language learning (see Table 3.10) and
ends up with the following claim:

Feature LI acquisition L2 (foreign language)

acquisition

Overall success . Children normally achieve Adult L2 learners are very

perfect mastery of their LI.
General failure Success is guaranteed.
Variation There is little variation among
LI learners with regard to
overall success or the path they
follow.
Goals The goal is target language
competence.
Fossilization Fossilization is unknown in child
language development.
Intuitions Children develop clear
intuitions regarding what is a
correct and incorrect sentence.
Instruction Children do not need formal
lessons to learn their LI.
Negative evidence Children’s ‘errors’ are not
typically corrected; correction
not necessary for acquisition.
Affective factors Success is not influenced by
personality, motivation,
attitudes, etc.
unlikely to achieve perfect
mastery.
Complete success is very rare.
L2 learners vary in both their
degree of success and the path
they follow.
L2 learners may be content with
less than target language
competence and may also be
more concerned with fluency
than accuracy.
L2 learners often cease to
develop and also backslide (i.e.
return to earlier stages of
development).
L2 learners are often unable to
form clear grammaticality
judgements.
There is a wide belief that
instruction helps L2 learners.
Correction generally viewed as
helpful and, by some, as
necessary.
Affective factors play a major
role in determining proficiency.

Table 3 10: Differences between LI and L2 acquisition (based on Bley-Vroman

1988)
108 The description of learner language

These general characteristics of foreign language learning tend to the con¬

clusions that the domain-specific language acquisition of children ceases to
operate in adults, and in addition, that foreign language acquisition re¬
sembles general adult learning in fields for which no domain-specific learn¬
ing system is believed to exist, (page 25)

Ervin-Tripp (1974), however, emphasizes the similarities she found in natur¬

alistic child learners of L2 French and LI learners:

We found that the functions of early sentences, and their form, their se¬
mantic redundancy, their reliance on ease of short-term memory, their
overgeneralization of lexical forms, their use of simple order strategies
were similar to processes we have seen in first language acquisition. In
broad outlines, then, the conclusion is tenable that first and second lan¬
guage learning is similar in natural situations (cited in Hatch 1978a: 205).

These two quotations reveal why researchers have reached such different
conclusions. Whereas Bley-Vroman concentrates on the product of acquisi¬
tion, Ervin-Tripp focuses on the process. Whereas Bley-Vroman talks about
foreign language learners, Ervin-Tripp refers to naturalistic learners.
L2 learners appear to tackle the problem of learning a language in similar
ways to LI learners. These similarities are most clearly evident in informal
learning situations when learners are attempting to engage in unplanned lan¬
guage use. But there are also differences in the ways in which L2 learners go
about ‘cracking the code5, and these become most evident in formal learning
situations. The differences between the kinds of learning involved in these
two settings are described by McNamara (1973) and d’Anglejan (1978). For
example, informal learning typically takes place in contexts where the input is
not consciously structured and the primary focus is on message conveyance,
while formal learning occurs in contexts where the input is usually carefully
organized and the primary focus is on form. Informal learning involves impli¬
cit knowledge, while formal learning is likely to involve at least some explicit
knowledge of L2 rules.
Formal and informal learning can also be differentiated in the kind of mem¬
ory learners rely on. Adult L2 learners have access to a more developed mem¬
ory capacity than LI learners and when they can use it (or are required to use
it, as in many pedagogic learning activities), differences between the language
they produce and that produced by LI learners occur. However, when they
are not able to use it, they will produce language that resembles young chil¬
dren’s. Cook (1977) found that when adults were unable to utilize their mem¬
ory capacity to process relative clauses they behaved in the same way as LI
learners. In another experiment designed to establish the number of digits
adult learners could remember in the L2, Cook found that they behaved like
native-speaking adults. This led Cook to conclude that when the memory
process depends on features of syntax, the same restrictions apply to the LI
 Developmental patterns: order and sequence 109

and adult L2 learner, but where the memory process is minimally dependent
on language, the adult L2 learner exploits his or her general memory capacity.
In other words, when L2 learners make use of their special language-learning
faculties, the identity hypothesis receives support, but when they rely on cog¬
nitive learning procedures of a general kind it does not.
Another obvious source of difference between LI and L2 acquisition lies in
the fact that L2 learners have access to a previously acquired language, in
some cases to several. There is clear evidence to show that this results in dif¬
ferences between L2 and LI acquisition—for example, in the case of the ac¬
quisition of German word order rules. Chapter 8 considers the various ways
in which the learner’s LI influences L2 acquisition.
The L2 = LI acquisition hypothesis is important because it raises so many
key issues. The evidence that we have considered here suggests that the hypo¬
thesis is partially supported. Given the immense cognitive and affective differ¬
ences between very young children and adults, the similarities in the language
they produce are striking. However, there are also significant differences
which have been shown to exist and, as we will see in Chapter 10, when we
consider whether L2 learners have continued access to Universal Grammar
and, also in Chapter 11, when we examine the role of age in L2 learning, these
differences have implications for theory-building.

Summary

The investigation of developmental patterns in learner language was motiv¬

ated by the desire to describe learner language in its own right, as a system of
rules that learners constructed and repeatedly revised. It has proceeded by
means of cross-sectional, longitudinal, and more recently experimental re¬
search. It has employed a number of methodological procedures, including
obligatory occasion analysis, target-like use analysis, frequency analysis, and
implicational scaling.
First language acquisition researchers sought evidence with which to test
the rival behaviourist and mentalist accounts of language learning. They pro¬
vided evidence of (1) an order of acquisition for morphemes, (2) sequences of
development in the acquisition of specific structures such as negatives, (3) an
overall pattern of development embracing grammatical, semantic, and prag¬
matic features of the TL, and (4) cross-linguistic similarities.
In L2 acquisition, early learner language is characterized by (1) a silent
period (although not in all learners), (2) extensive use of formulas and (3)
structural and semantic simplification, particularly in unplanned language
use. Early studies of the order of acquisition of grammatical morphemes such
as plural -s and articles produced mixed results. The cross-sectional studies
reported a consistent accuracy order when the learner was focused on com¬
municating meaning, which some researchers equated with acquisition order.
Krashen (1977) proposed a ‘natural order’ of acquisition based on these
110 The description of learner language

studies. However, longitudinal studies did not always confirm this order. A
major criticism of the morpheme studies is that acquisition is viewed as one of
‘accumulated entities’, and insufficient attention is paid to the ways in which
learners achieve gradual mastery over specific linguistic features. However,
studies of individual morphemes such as pronouns, and syntactical structures
such as negation, lend strong support to the existence of developmental se¬
quences. Learners of different L2s manifest similar patterns of development
when acquiring pronouns. For example, they all omit pronouns and/or re¬
place them with nouns, and they all overgeneralize individual pronouns.
Also, certain semantic features (for example, person) appear to be acquired
before others (for example, gender). In the case of syntactic structures the
regularities are even more evident. The acquisition of English and German
negation involves a series of transitional stages in which learners gradually
switch from external to internal negation and from pre-verbal to post-verbal
negation. Relative clauses constitute a complex learning task. There is evid¬
ence to suggest that learners solve it piecemeal by learning first to modify
noun phrases before the verb, and then noun phrases that follow the verb.
Also, learners acquire the functions that relative pronouns can perform in a
fairly well-defined order. Finally, the ZISA project has provided some im¬
pressive evidence to show that learners acquire German word order rules in a
clear sequence. This work has since been further extended to cover the ac¬
quisition of a range of features in L2 English. ZISA researchers have also dis¬
tinguished developmental and variational features.
A comparison of the developmental patterns found in LI and L2 acquisi¬
tion lends partial support to the identity hypothesis. Some striking similarities
have been found in syntactic structures such as negatives, but there are also
differences. It has been suggested that adult L2 learners are more likely to
manifest similar patterns of acquisition to children acquiring their native lan¬
guage when they engage in informal learning, but that they also have access to
formal learning strategies, which result in differences.
#

Conclusion

In the Introduction to an influential collection of papers, Hatch (1978a)

wrote:

Nothing can be certain until second language acquisition has been studied
in tangible case histories or until empirical evidence has been obtained
(1978: 10).10

A considerable amount of empirical evidence is now available. This chapter

has reviewed a selection of it in order to examine to what extent there are tes-
tifiable developmental patterns (orders and sequences) in L2 acquisition of
the kind that have been observed in LI acquisition.
 Developmental patterns: order and sequence 111

What constitutes evidence for a developmental pattern? This question has

rarely been addressed in an explicit manner, but the following would seem to
be the criteria that researchers have applied:

1 Developmental patterns can be established by looking at either the order

in which different target structures are acquired, or the sequence of
stages through which a learner passes en route to mastery of a single TL
structure.
2 In the case of transitional structures, a ‘stage’ consists of a period during
which learners use a particular form or structure in a systematic manner,
although not necessarily to the exclusion of other forms and structures.
3 The forms and structures that a learner produces at different points during
the process of L2 acquisition can be ordered such that one form or struc¬
ture always precedes another.
4 Learners progress step-by-step along an order or a sequence, mastering
one particular structure—target-language or transitional—before
another.
5 Strong evidence for developmental patterns occurs when it can be shown
that an order or a sequence is universal (i.e. applies to different L2s and
to all learners). Weaker evidence is found if it is shown that an order or
a sequence applies only to specific L2s and/or to specific groups of
learners.

These criteria raise many questions to do with the definition of such key
terms as ‘form’, ‘structure’, ‘systematic use’, and ‘master’. Different re¬
searchers have worked with rather different operational definitions, with the
result that it is not easy to compare results across studies. Inevitably, then, the
picture is often a fuzzy one, suggesting that it may be better in many cases to
talk of ‘regularities’ than of ‘definite patterns’.
The strongest evidence for developmental patterns undoubtedly comes
from studies of syntactic structures. We have seen that the morpheme studies,
on which great store was once set, are of doubtful validity, because their view
of acquisition as one of ‘accumulated entities’ is seriously flawed. But in the
case of negatives, relative clauses, and word order rules (as well as other areas
of syntax such as interrogatives), there is evidence to support at least the
‘weak’ definition of developmental patterns referred to in (5) above. The ac¬
quisition of these structures in a particular L2 shows surprising uniformity,
and in some cases the regularities hold across different L2s. For example,
learners of all L2s so far investigated appear to go through an initial stage of
preverbal negation.
There is also plenty of evidence that acquisitional sequences are not com¬
pletely rigid. For example, in the case of pronouns all learners seem to go
through a stage when they overgeneralize one pronominal form, but they may
vary in the form they choose to overgeneralize. A sequence can also be influ¬
enced by the learner’s LI. Thus, the starting point for the acquisition of
112 The description of learner language

German word order rules seems to vary according to the basic word order of
the learner’s LI. Also, the LI may result in an additional stage, as when Ger¬
man and Norwegian learners of L2 English add a negator after the main verb
at a certain developmental point.
It is also necessary to recognize the limitations of the research which has
investigated developmental patterns.

1 Focus on grammar
As was pointed out in Chapter 1, a major limiting factor in SLA research has
been its preoccupation with grammar. This chapter has concerned itself only
with research into the acquisition of grammar. It is worth briefly noting, how¬
ever, that the idea of developmental sequences has also been explored in L2
phonology. It has been claimed, for example, that learners have a marked
preference for open syllables (i.e. syllables that end in a vowel) as opposed to
closed syllables (i.e. syllables that end in a consonant) in the early stages
(Tarone 1978). This is reflected in a universal tendency towards epenthesis
(the insertion of a vowel where it is not required in the TL) and consonant de¬
letion.11 It also appears to be the case that whereas learners often substitute
LI sounds for TL sounds, this does not always happen. Wode (1977), for in¬
stance, found that the German children he studied followed the same devel¬
opmental sequence for /r/ as that observed in native, English-speaking
children. However, as Ioup and Weinberger (1987) note, it is still not clear
why developmental processes sometimes occur and sometimes do not in L2
phonology.12
There has also been some work on developmental patterns in L2 lexis,
which led Meara (1984) to speculate that learners go through transitional
stages. Meara’s own work on learners’ word associations concentrated on
showing the qualitative differences in the networks of word associations be¬
tween learners and native speakers. Learners’ responses tend to be very varied
and unpredictable when they are given a stimulus word and asked to produce
an association, which ‘makes it very difficult to use word associations as an
index of development, or to link change in association patterns with progress
in other areas’ (1984: 232). Even allowing for Sharwood Smith’s (1984) re¬
joinder that Meara underestimates what his own research shows, it is prob¬
ably premature to speak of anything so defined as an ‘order’ or a ‘sequence’ in
the L2 acquisition of lexis.
Ideally, the study of L2 vocabulary acquisition requires longitudinal stu¬
dies, but there have been very few of these, most researchers preferring, like
Meara, to conduct experimental, cross-sectional studies. Two exceptions to
this generalization are Yoshida (1978) and Wode et al. (1992). Yoshida in¬
vestigated the acquisition of English by a young Japanese child. One finding
was that this child showed a marked propensity for nouns over verbs in the
early stages of development. Wode et al. report the results of their study of the
 Developmental patterns: order and sequence 113

English vocabulary acquired by four German children in a natural setting.

They point to three differences between LI and L2 vocabulary acquisition.
Whereas vocabulary growth is slow in LI acquisition up to the first fifty
words and then rapidly accelerates, it is initially much more rapid but soon
decelerates in L2 acquisition. The kind of overgeneralizations evident in LI
acquisition (for example, ‘cat’ is used to refer to a variety of four-legged an¬
imals) is not apparent in L2 acquisition. Closed-class items like prepositions,
articles and pronouns are acquired late in LI children, while L2 learners
manifest a high percentage of such items early on. Wode et al. explain these
differences in terms of the cognitive abilities and background knowledge of
the two types of learner.
Clearly, it is not yet possible to say much about developmental patterns in
L2 vocabulary acquisition. The research to date, however, suggests that such
patterns may exist.

2 Lack of a general index of second language acquisition

We noted that in LI acquisition research, mean length of utterance (MLU)
served as a reliable measure of a child’s development, at least until the later
stages. However, as Larsen-Lreeman (1978) observes, MLU cannot be used
in SLA research because so many of the learners’ early utterances are formu¬
laic. Various alternative devices have been suggested, but none has become
widely accepted. The result is that it is difficult to make reliable comparisons
between learners.

3 Inter-learner variability
If the variability between learners becomes too great, it makes little sense to
continue to talk of a standard developmental route, a point raised by Light-
bown (1984). She claims that for every study that reports an order or a se¬
quence, there is another study which has produced counter-evidence and
argues that the learners’ LI, the input they experience, and their socio-
psychological attitudes can all result in variant patterns. But Lightbown’s
conclusion is not that clear patterns do not exist, merely that we need to ex¬
amine how these variables interact with universal tendencies. Again, this
raises the whole question as to what constitutes a ‘definite pattern’ of L2 ac¬
quisition, a question partly but not completely answered by the criteria listed
above.

4 Intra-learner variability
The learner language produced by a single learner is likely to display marked
variability. Thus, at any one stage of development, a learner is likely to
114 The description of learner language

produce utterances reflecting different stages of development. How then can

we speak of a clear developmental pattern? The answer that researchers like
Cancino et al. (1978) have given is to show that it is possible to identify se¬
quences in terms of gradual shifts in the frequencies of the varied structures a
learner employs at different stages. However, if the variability becomes too
great it makes little sense to talk of a ‘stage’ of development.
Intra-learner variability also poses problems of description. Bley-Vroman
(1983) has pointed out that nobody to date has succeeded in writing a ‘sys¬
tematic grammar’ of learner language—even for L2 English, the most studied
language. The descriptions that do exist are piecemeal. The closest to an over¬
all account of the developmental path is that provided by Johnston and
Pienemann (1987), but this still only accounts for a small proportion of the
total grammar of a language.

5 Methodological problems
It is easy to find methodological flaws with many of the studies. Lightbown
(1984) lists a number of them. Some studies have been carried out in a manner
that makes replication of them difficult and yet, clearly, replication is essen¬
tial if claims about the universality of sequences are to be substantiated. An¬
other problem is that researchers have tended to base their claims on data
collected by means of a single instrument and, thus, have ignored the variabil¬
ity in learner language produced in different contexts. Lightbown also sug¬
gests that some researchers have used inappropriate data collection
procedures for their subjects, while others have presented the results anec¬
dotally rather than quantitatively. The need for quantification is not some¬
thing that all researchers would acknowledge, however.13
Despite all these problems, the work on developmental patterns has been
substantial and in many cases convincing. But how significant is this work? In
the opinion of some, it is not very important. Sharwood Smith (1984) argues
that searching for developmental sequences is less important than revealing
the qualitative ways in which the system a learner builds at a given time differs
from that of the native speaker. Others, like Bley-Vroman (1983), have ar¬
gued that learner language must be described in its own right without refer¬
ence to the TL. This latter view has underscored some of the best descriptive
work in SLA research.
The discovery that learners do follow identifiable routes has generated im¬
mense interest. Any theory of L2 acquisition will need to account for develop¬
mental patterns. The theory that has been dominant in SLA—interlanguage
theory (Selinker 1972)—was initially formulated to provide such an account.
It claimed that learners construct a series of interlanguages (i.e. mental gram¬
mars that are drawn upon in producing and comprehending sentences in the
L2) and that they revise these grammars in systematic and predictable ways as
they pass along an interlanguage continuum. It has also proposed that this
 Developmental patterns: order and sequence 115

continuum involves both recreation and restructuring (Corder 1978a).

Learners create unique rules not to be found in either the LI or the TL and
then gradually complexify these rules in the direction of those in the TL, par¬
ticularly in the case of syntax learnt in informal environments. They also
make use of their LI, gradually restructuring it as they discover how it differs
from the TL, a process that Corder suggests is more prevalent in the case of
phonology learnt in formal environments, a point taken up in Chapter 8.
Thus, both universal principles and the learner’s LI are involved in the pro¬
cess of acquiring an L2. How these work and interact is of crucial importance
for understanding L2 acquisition and is the main subject of Section Three of
this book.

Notes

1 In fact, though, many of the subjects of the studies reported in this chapter
were of the ‘mixed’ kind. That is, although they had opportunities for ex¬
posure in non-instructional settings they were also receiving some class¬
room instruction. This is particularly true of children learning the L2 in a
country where it functions as an official language (see, for example, Dulay
and Burt 1973; Pienemann 1980).
2 A second problem with target-language based analyses is that of deciding
which set of TL norms to use as the basis for the comparison with learner
language. These norms vary somewhat in different dialects of the TL and
it cannot be assumed that all L2 learners are targeted on the standard dia¬
lect. This problem is further discussed in Chapters 5 and 6.
3 The label SHARP is not an acronym, in fact. Atkinson chose it as an ant¬
onym to FLAT, the other tradition followed by LI researchers. This is an
acronym, referring to ‘First Language Acquisition Theories’.
4 According to one view of L2 acquisition, children begin with a non-
linguistic system of representing objects and their relationships (for ex¬
ample, a sensory motor representation) and then learn how this system
‘maps’ on to the linguistic system (see Sinclair-de-Zwart 1973).
5 There is also the general problem of determining exactly what a learner’s
‘silence’ signifies. Silence can be used as a conversational strategy—a
means of indicating disagreement, disapproval of someone’s behaviour,
or plain uncooperativeness, for example.
6 The fact that structural/semantic simplification can reflect the processes
either of acquisition or production raises an epistemological and meth¬
odological problem in SLA research. What actually counts as ‘acquisi¬
tion’? To what extent is it possible to distinguish between what learners
‘know’ and what they ‘do’, given that the primary evidence for what they
know rests in what they do? This issue raises its head at various points
throughout this book and is discussed in some detail in Chapter 15.
116 The description of learner language

7 The acquisition of pronouns is further considered in Chapter 10 when

studies working within the framework of Universal Grammar are
discussed.
8 The retention of pronominal copies may be a manifestation of the same
psycholinguistic mechanism underlying the choice of nouns over pro¬
nouns—the learner’s attempt to be transparent by ensuring that all mean¬
ings map on to overt forms.
9 The claim that variational features like copula are not acquired in a fixed
order does not preclude their acquisition involving a sequence of stages.
Indeed, there is some evidence to suggest that the acquisition of copula
‘be’ does involve transitional stages (see Ellis 1988a and Chapter 4).
Thus, copula may first appear in learner language at any time; its mastery,
however, takes time and involves stages of development.
10 It is assumed that Hatch’s use of ‘or’ is not meant to suggest that ‘case
studies’ do not constitute ‘empirical research’. They clearly do.
11 Not all researchers agree with Tarone that there is a universal tendency
towards open syllables in interlanguage. Sato (1987), for example, found
negligible use of epenthesis in her Vietnamese learners of English, which
she saw as ‘evidence against the hypothesized universal preference for the
CV (i.e. open) syllable’ (1987: 260). Riney (1990) provides evidence to
suggest that the crucial factor may be the age the learner began learning
the L2 (see Chapter 11).
12 Much of the work on L2 phonology focuses on the influence of the learn¬
er’s LI and for this reason is considered in greater detail in Chapter 8,
when language transfer is considered. Useful overviews of L2 phonology
research can be found in Leather and James (1991).
13 There is a rich ethnographic and descriptive tradition in SLA research
which eschews quantification in favour of rich, illustrative accounts of L2
learners (see Johnson 1992, in particular Chapters 4 and 6, for an account
of these).

Further Reading

A good account of the methodological procedures used in the description of

learner language can be found in:
M. Long and C. Sato, ‘Methodological issues in interlanguage studies: an in-
teractionist perspective’ in A. Davies et al. (eds.), Interlanguage (Edinburgh
University Press, 1984).
A very readable general account of the acquisition of English as an LI is
Chapter 2 of G. Wells, The Meaning Makers (Hodder and Stoughton, 1986).
More detail is available in the articles in P. Lletcher and M. Garman (eds.)
Language Acquisition (Cambridge University Press, 1986).
An interesting study of the role of the silent period in L2 acquisition can be
found in M. Saville-Troike, ‘Private speech: evidence for second language
 Developmental patterns: order and sequence 117

learning strategies during the ‘silent’ period.’ Journal of Child Language

(1988)15:567-90.
An informative study of formulaic speech in L2 acquisition is
L. Rescorla and S. Okuda, ‘Modular patterns in second language acquisi¬
tion’, Applied Psycholinguistics (1987) 8: 281-308.
There is now a wealth of descriptive studies of L2 acquisition to draw on,
which makes selection very difficult. Key volumes providing collections of
some of the most important papers are:
E. Hatch (ed.), Second Language Acquisition (Newbury House, 1978).
R. Andersen (ed.), The Acquisition and Use of Spanish and English as First
and Second Languages (TESOL, 1979).
S. Felix (ed.), Second Language Development (Gunter Narr, 1980).
R. Scarcella and S. Krashen (eds.), Research in Second Language Acquisition
(Newbury House, 1980).
R. Andersen (ed.), Second languages: A Cross-linguistic Perspective (New¬
bury House, 1984).
G. Ioup and S. Weinberger (eds.), Interlanguage Phonology: The Acquisition
of a Second Language Sound System (Newbury House, 1987).
Articles by E. Hatch and S. Corder in J. Richards (ed.), Understanding
Second and Foreign Fanguage Learning (Newbury House, 1978), provide
useful general accounts of learner language.
Many of the key articles have appeared in journals—in particular, In¬
terlanguage Studies Bulletin, Working Papers on Bilingualism, Language
Learning, TESOL Quarterly, and Studies in Second Language Acquisition.
The first two are no longer published, but are available in many university
libraries.
V

*
4 Variability in learner language

Introduction

In Chapters 2 and 3, evidence was provided to suggest that learner language

displays significant systematicity. However, we also noted that a pervasive
feature of learner language is its variability. We have seen that learners some¬
times make an error in the use of a specific target-language structure, and
sometimes do not. Also, we saw that the identification of a ‘stage’ of develop¬
ment in the sequence of acquisition does not mean that learners consistently
make use of a single form or pattern, but rather that they show a preference
for the use of one form among others that they use during the same period.
Any researcher who has grappled with L2 data has had to recognize and find a
way of dealing with variation.
There are, in fact, three approaches to resolving the apparent contradiction
between variation and systematicity in learner language. The first approach is
that practised by linguists in the Chomskyan tradition (for example, White
1989a and Gregg 1989), who adopt what Tarone (1983) has called ‘a homo¬
geneous competence model’. In this approach variation is seen as a feature of
performance rather than of the learner’s underlying knowledge system. The
general claim is that in order to study language it is necessary to abstract what
learners ‘know’ from what they ‘do’. This involves various kinds of idealiza¬
tion. Lyons (1972) proposes three ways in which abstract ‘sentences’ can be
derived from actual ‘utterances’: (1) regularization, the elimination of vari¬
ation related to the speech disfluencies that occur in natural language use, (2)
standardization, the elimination of the variation found in different dialects of
a language, and (3) decontextualization, the elimination of the variation asso¬
ciated with the use of language in different social situations. As a result of
these idealizing processes, the linguist gains access to data that are invariable
and so can be used to investigate the learner’s linguistic competence. The type
of data often preferred by researchers who operate within the homogeneous
competence paradigm consist of speakers’ intuitions regarding what they
think is correct in the L2 rather than actual instances of language use.1 The
problem of variability, therefore, is solved by distinguishing ‘competence’
and ‘performance’, and by establishing the description and explanation of
linguistic competence as the main goals of enquiry. In effect, then, variability
is ignored in this paradigm.
120 The description of learner language

The second approach is a sociolinguistic one; its goal is to study language in

relation to social context. There are a number of sociolinguistic models that
have informed SLA research (see Beebe 1988b). These seek to account for the
variation evident in different varieties of English (for example, from one dia¬
lect to another), the variation which occurs among speakers who differ in
terms of general social factors such as class and ethnicity, and also the vari¬
ation which arises within the speech of a single speaker as a result of changes
in situational context. Researchers in this tradition (for example, Beebe 1980;
Dickerson 1975) aim to describe both the learner’s linguistic and socio¬
linguistic competence. They are concerned, therefore, both with what learn¬
ers know about the L2 system and with what they know about how it is used
in communication. Not surprisingly, they prefer to work with data that re¬
flect actual instances of language use rather than with intuitions. In this ap¬
proach, the problem of variability is solved by demonstrating that it is
patterned—in other words, that language use can be both variable and
systematic.
The third approach is psycholinguistic. Psycholinguistic processing models
seek to account for the variation that results from factors that influence the
learner’s ability to process L2 knowledge under different conditions of use.
For example, systematic differences in performance have been found to exist
in learner language depending on whether it is planned or unplanned (Ellis
1987b; Crookes 1989). Researchers in this tradition also make use of data re¬
flecting natural language use. They seek to show that planning variability is
systematic and to explain it in terms of mental processing.
Thus, whereas the linguistic approach ignores variability, the sociolinguis¬
tic and the psycholinguistic approaches try to describe and explain it. Socio¬
linguistic approaches treat social factors as primary, although, as we will see,
they may also refer to psycholinguistic mechanisms to explain how situ¬
ational factors result in variability. Psycholinguistic approaches are con¬
cerned with identifying the internal mechanisms responsible for variable
performance and pay little attention to the social factors that motivate them.
Sociolinguistic and psycholinguistic approaches, therefore, are complement¬
ary and a full account of variability in learner language requires both.
L2 variability has attracted increasing attention in SLA research, drawing
on a range of different theoretical perspectives. Wolfram (1991: 104), in a re¬
view of a number of books on L2 variability, finds the ‘range of models and
perspectives almost overwhelming in its inclusiveness’ and considers the
‘search for a unitary model of SLA variation very elusive’. Similarly, Zuengler
argues that ‘it is misguided to search for one comprehensive theory, since one
theory will most likely be insufficient in explaining the complexity of per¬
formance variation ...’ (1989: 66). The aim of this chapter is not to provide a
‘unitary model’ but rather to concentrate on the descriptive information
which has been uncovered.
 Variability in learner language 121

Finally, it needs to be pointed out that because learner language is a natural

language, the types of variability that it manifests are not unique. All natural
languages will manifest variable forms that have two or more variants. A
variable form is a feature (phonological, lexical or grammatical) that is real¬
ized linguistically in more than one way. Variants are the linguistic devices
that realize a variable form. For example, in native speaker English, copula
‘be’ constitutes a variable form; it has two variants (full copula and con¬
tracted copula). Of course, the variants found in learner language will not
necessarily be the same as those found in the target language. Thus, learners’
interlanguages have been found to manifest three rather than two variants for
copula ‘be’ (zero, full, and contracted). Also, as we have already noted, vari¬
ability seems especially prevalent in learner language. However, the types of
variability and their sources are the same as those observed in natural
languages.
The starting point for this chapter is a fuller discussion of the socio-
linguistic and psycholinguistic theories that have informed variability
research in SLA research. A taxonomy of the different types of variability in
learner language follows. This provides a basis for an examination of the em¬
pirical research into L2 acquisition. Finally, further consideration will be
given to the contradictory views regarding the significance of variability
phenomena in L2 acquisition.

Some theoretical perspectives

A number of different lines of primarily sociolinguistic enquiry have in¬
formed the study of variability in learner language. These include (1) the
Labovian paradigm, (2) the dynamic paradigm, and (3) the social psycholo¬
gical perspective of Accommodation Theory. This section will also examine
psycholinguistic models of language production and, in particular, the dis¬
tinction between planned and unplanned discourse, as this has also been used
to account for L2 variation. These models are presented, in the first instance,
without reference to how they have been exploited in SLA research.

Sociolinguistic models

The Labovian paradigm

The Labovian paradigm has exerted the greatest influence on the study of
variability in SLA research, particularly in much of the earlier work. Two
constructs are of particular importance: speech styles and variable rules.
Labov distinguishes social factors (responsible for inter-speaker variation)
and stylistic factors (responsible for intra-speaker variation). Social factors
such as social class, age, and gender can account for differences among
122 The description of learner language

speakers. The variable forms in which these individual differences are evident
are called indicators. An example of such an indicator is -ing, which may be
pronounced [in] in a number of dialects in accordance with such social factors
as sex and social class (see Fischer 1958). According to Labov and other
sociolinguists, some sociolinguistic variables—such as -ing—are ‘more
highly developed’ in the sense that they are both ‘social’ and ‘stylistic’ (i.e.
manifest both inter-speaker and intra-speaker or stylistic variation). They are
referred to as markers. Bell (1984), in fact, has argued that there is no clear
distinction to be made between social and stylistic variation.
Labov (1970) lists five axioms relating to the study of speech styles (i.e.
markers):

1 ‘... there are no single style speakers’. All speakers vary their language to
some degree when the social context or topic changes.
2 ‘Styles can be ranged along a single dimension, measured by the amount of
attention paid to speech’. Language users vary in the degree to which they
monitor their speech in different situations.
3 The vernacular style is the style in which minimum attention is given to
monitoring speech. It is the style associated with informal, everyday speech
and it provides ‘the most systematic data’ for linguistic study.
4 It is not possible to tap the vernacular style of users by systematic observa¬
tion of how they perform in a formal context (such as an experiment).
5 The only way to obtain good data on the speech of language users is
through systematic observation.

The conflict between the fourth and the fifth axioms leads to what Labov
calls the Observer’s Paradox. Good data require systematic observation, but
this prevents access to the user’s vernacular style.
As an example of how Labov set about examining speech styles, let us con¬
sider one of his studies. Labov (1970) examined the speech patterns of New
Yorkers. He collected data using a variety of tasks in order to sample a range
of speech styles, which he classified as (1) casual speech (i.e. the relaxed
speech found in the street and in bars), (2) careful speech (for example, the
speech found in interviews), (3) reading, (4) word lists, and (5) minimal pairs.
These styles were spread along a continuum according to the amount of at¬
tention paid by the speakers to their own speech, the least attention being
paid in (1) and the most in (5). Thus, attention is seen as the mechanism
through which other factors can affect style. Labov’s model, therefore, al¬
though primarily sociolinguistic, also incorporates a psycholinguistic fac¬
tor—attention. Attention serves as the mechanism through which causative
social factors such as verbal task (in particular), topic, interlocutor, setting, or
the roles of the participants influence actual performance.
Labov investigated a number of pronunciation features and was able to
show that the use of sounds like /0/ (i.e. the first sound in ‘thing’) and their
 Variability in learner language 123

variants (for example, /t/) functioned as markers. Speakers used the prestige
/0/ more frequently in styles where they were able to pay attention and the less
prestigious sounds such as Itl in styles where little or no attention to speech
was paid. Labov referred to these changes in speech as style shifting.
Labov’s work indicated that style shifting was systematic and, therefore,
predictable. This systematicity can be of two kinds: categorical and prob¬
abilistic. Categorical systematicity is evident when it can be shown that
speakers always use one particular feature (such as /0/) in one style and an
other (such as /t/) in a different style. In such cases it is possible to write a cat¬
egorical rule to describe the speech behaviour. Such a rule has this form:

X-»Y/_A

where X refers to the variable itself, Y to its actual realization, and A the par¬
ticular context or style (for example, the first sound of ‘thing’ is realized as Itl
in a casual style). The actual behaviour of Labov’s subjects, however, was not
usually categorical in this way. They tended to use one variant in one style and
another variant in another style to a greater or lesser extent. In other words,
their behaviour was probabilistic. To account for this, Labov proposed the
use of variable rules. These have a much more complex form to account for
the probability of occurrence of a specific feature in a given style or context.
Fasrch (1980) provides the following example:2

Y with the probability of .6 in context/style A

Z with the probability of .4
X
Y with the probability of .9 in context/style B
Z with the probability of .1

This states that a given variable feature, X, is manifest as either Y or Z with

differing levels of probability depending on the context/style. Such a rule can
account for the patterns of variability in the choice of /0/ and Itl which Labov
found in the speech of New Yorkers. It can show that speakers are much more
likely to use the prestige feature, Y, in a careful style, B, than in a more casual
style, A, and, conversely, that they are less likely to use the less socially presti¬
gious feature, Z, in B than in A.
Variable rules have been used to describe the extent of the systematic vari¬
ation that occurs in relation to situational factors (i.e. in style shifting) and
also that which arises as a result of linguistic context. For example, Labov
(1969a) was also able to show that the use of variants of copula ‘be’ (full, con¬
tracted, and zero copula) in Black English Vernacular (BEV) was influenced
by the preceding and following elements in the sentence. Thus, zero copula
was most likely to occur when the preceding word ended in a vowel and
‘gonna’ followed:

He gon’ try to get up. (+ vowel/+ verb)

124 The description of learner language

and least likely to occur when the preceding word ended in a consonant and a
noun phrase followed:

Bud is my friend. (+ consonant/+ noun phrase)

A variable rule can express the probability of a particular form being used in a
particular linguistic context.
More recently, powerful statistical programmes such as VARBRUL have
been developed to account for the effects that various factors relating to both
situational and linguistic context can have on speakers’ choice of language
forms (see Preston 1989: 14ff for an account of this procedure). VARBRUL
provides a means of determining the differential effect of a number of factors
and, also, how they interact.
A question of considerable theoretical importance is the status of variable
rules. Labov argues that the variation described through variable rules is ‘an
inherent and regular property of the system’ (1970: 225) and that it cannot be
accounted for in terms of dialect mixing and switching. Furthermore, Labov
dismisses claims that variable rules reflect ‘performance’ rather than ‘compet¬
ence’, even though he accepts that they are rules of production rather than
reception. He argues that there is ‘a deep asymmetry between perception and
production’ (1970: 226). For Labov, then, variable rules are an aspect of that
competence that underlies production. He claims that it can be studied only
through examining language in use, as speakers have no intuitive awareness
of such rules.
The Labovian paradigm has come in for considerable criticism in recent
years (for example, Wolfson 1976; Beebe 1982; Bell 1984; Rampton 1987).
One of the main problems consists of Labov’s use of ‘attention to speech’ as a
causative factor in style shifting. As Wolfson points out, very little is known
about the relationship between audio-monitoring and speech style and it is
difficult to obtain independent measures of attention. Thus we cannot be sure
that a speaker really does attend to speech more in a ‘careful’ than in a ‘casual’
style. Sometimes speakers may pay close attention to a feature in an informal
setting, for example, if the feature in question is important for communicat¬
ing in this kind of language use or, as Hulstijn (1989a) suggests, simply
because the learner lacks full control of a feature. A further serious criticism,
advanced by Bell (1984), is that the attention-to-speech model ignores the
effect that the addressee has on the interaction. It cannot account for the role
played by such factors as feelings of ethnic identity and solidarity with one’s
in-group. Beebe argues that ‘attention to speech is inadequate as an explana¬
tion for style’ (1982: 13) and points to a number of studies (for example,
Bourhis and Giles 1977) where style shifting does not appear to involve atten¬
tion on the part of speakers. Beebe also criticizes Labov’s preoccupation with
the standardness dimension of style shifting (i.e. whether speakers use a
standard or non-standard linguistic norm), pointing out that ‘shifts have been
found in amount of talk, degree of elaboration, speech rate, duration, vocal
 Variability in learner language 125

intensity, pause and utterance length, stress, pitch, intonation, content ex¬
pressed and even complete code switches’ (page 15). Variable rules have also
been attacked. In particular, doubts have been expressed as to whether the
probabilities expressed in a variable rule can be seen as part of the native
speaker’s competence, as Labov claims. As we will see in the next section,
Bickerton (1975) has shown how variability can be accounted for entirely
through categorical rules. Variable rules, therefore, may be unnecessary.
Wolfram (1991), in fact, dismisses them as ‘passe’. However, other socio-
linguists such as Preston (1989) continue to see value in them and, more espe¬
cially, in VARBRUL.3
The Labovian paradigm has had a profound impact on the study of variab¬
ility in learner language. It provided the methodology for a number of early
studies (for example, Dickerson 1975 and Beebe 1980) and also served as the
basis for theories of L2 acquisition (for example, Tarone 1982; 1983).

The dynamic paradigm

Whereas Labov sought to account for variability in groups of speakers by

means of variable rules, Bailey (1973) and, in particular, Bickerton (1975),
developed alternative means, based on a view of how language change takes
place.
Bailey sought to show how a theory of language change can account for
synchronic variability in language use. According to his Wave Theory, lin¬
guistic innovation is first introduced by one group of speakers. By the time it is
taken up by a second group, the first group has introduced a second innova¬
tion. And so as old rules spread, new rules arise. The spread, or diffusion, of
new rules also takes place in another way. Initially, a rule may be restricted to
a specific linguistic environment and then gradually come to be used in an
increasing range of environments. Linguistic environments can be distin¬
guished according to their weight. Change originates in ‘heavy environments’
(i.e. those environments that favour the use of a particular variant) and then
spreads through intermediate to ‘light environments’ (i.e. those environments
that do not favour the use of the variant). For example, for zero copula, the
heaviest context in Black English Vernacular is + vowel/ + gonna and the
lightest + consonant/ + noun phrase (see pages 123-4 for examples). Thus,
like Labov’s variable rule, the wave theory accounts for the systematic effects
of both social and linguistic factors.
The relationship between language change and synchronic variation (i.e.
the variation evident in speakers at a single point in time) is clearly evident in
creoles—languages that began as pidgins but then developed rapidly when
they came to be used and learnt as first languages. A creole generally has sev¬
eral varieties, or ‘lects’ as they are called: (1) the basilect manifests the ‘deep¬
est’ creole features, (2) the mesolect contains middle-level features, and (3)
the acrolect is the variety closest to the standard language. Together the lects
126 The description of learner language

constitute a creole continuum, going from the simplest, most basic variety to
the most complex. In a study of Guyanese Creole, Bickerton (1975) was able
to show how different speakers could be located at different points on the
continuum by using implicational scaling (see Table 4.1). Thus, speakers 1
and 2 (whose speech provided evidence of only one of the four grammatical
features Bickerton examined) were basilectal speakers, while speakers 3 and
4 (who provided evidence of two or three of the features) were mesolectal,
and speakers 5 and 6 (who used all four features) were acrolectal. Table 4.1
constitutes an implicational scale because it can be shown that the presence of
a feature to the left in the horizontal list of features entails the presence of all
features to the right. For example, an acrolectal speaker who has acquired
Ving will also have acquired Ning, doz, and a. Bickerton claims that it is pos¬
sible to write separate grammars consisting only of categorical rules to ac¬
count for each lect. The resulting description is called a polylectal grammar.

Speaker Linguistic features

Ving Ning doz a

1 0 0 0 X
2 0 0 0 X
3 0 0 X X
4 0 X X X
5 X X X X
6 X X X X

X = occurs; 0 = does not occur

Table 4.1: Varieties of Guyanese Creole in the speech of six speakers (simplified
table from Bickerton 1975: 79)

Different types of variability can be accounted for using Bickerton’s model.

First, there is inter-speaker variation. Some speakers may have access to one
variety and others to other varieties, although the full continuum of varieties
will only be evident in the ‘langue’ of the speech community as a whole (see
page 130 for a definition of langue). Second, there is intra-speaker variation.
This occurs when speakers who have access to more than one lect engage in
code-switching. In accordance with situational factors such as the topic of the
discourse, its purpose, and the addressee, a speaker may choose to use one
lect sometimes and another at other times. Code-switching works in much the
same way as style shifting, the difference being that in the dynamic paradigm
it can be accounted for in terms of a polylectal grammar rather than variable
rules. Third, there is free variation. When a speaker first acquires a new
feature, this is likely to exist alongside an existing feature and to be
used to realize the same meanings. Bickerton claims that this occurs only in
those speakers whose language systems are still unstable—a relatively small
 Variability in learner language 127

set of the total population. He also claims that free variation is short¬
lived.
Free variation becomes evident only when a form-function analysis is un¬
dertaken. In this kind of analysis, a specific form is chosen for study. All in¬
stances of this form are then identified in the data and the different meanings
which it realizes described. Free variation is held to occur when it can be
shown that two or more forms are used randomly to perform the same set of
functions. This kind of analysis is important in SLA research because it pro¬
vides a means of showing what Tarone (1988: 11) refers to as ‘internal vari¬
ation’ (i.e. variation in the learner’s L2 system that is independent of the
target language). In this respect it contrasts with error analysis and perform¬
ance analysis, which describe learner language in relation to how it differs
from the target language and, therefore, can only show ‘external variation’.
The dynamic paradigm offers the SLA researcher some powerful theoret¬
ical constructs (for example, the notion of environmental weight, of distinct
lects, and of code-switching), together with some useful tools for in¬
vestigating L2 variability (such as implicational scaling and form-function
analysis). However, it is not without its problems. It is not clear, for instance,
to what extent it is possible to identify distinct varieties in learner language as
Bickerton was able to do for Guyanese Creole. L2 acquisition involves even
more rapid change than creolization, with the result that the stages of devel¬
opment are not sharply defined. The idea of a polylectal grammar, while at¬
tractive, may not be applicable to L2 acquisition. Also, as Preston (1989)
notes, Bickerton’s dismissal of variable rules may not be warranted, given
that at least some creole speakers manifest a variable grammar—a point that
may be even more relevant in the case of L2 learners.
The dynamic paradigm has influenced a number of SLA researchers.
Gatbonton’s (1978) gradual diffusion model owes much to Bailey’s work.
Huebner (1979; 1983) and Tarone in her later work (for example, Tarone
1985) have found form-function analysis revealing.

Social psychological models

Social psychological models seek to relate the language attitudes of speakers
to their actual language use. In other words, they attempt to explain variation
in language use by reference to the speakers’ views about the social, institu¬
tional, and ethnic status of their own in-group, and also that of out-groups
with whom they come into contact. A number of social psychological models
have been used in SLA research (see Gardner 1985 and Beebe 1988b). How¬
ever, one model in particular has been used as a basis for examining L2 vari¬
ability—Giles’ Speech Accommodation Theory—and so we will focus on
this, reserving a more thorough consideration of social psychological ac¬
counts of L2 acquisition for Chapter 6, when we look in detail at the role
played by social factors.
128 The description of learner language

Speech accommodation theory identifies three principal types of variation,

according to the nature of the adjustments which speakers make to their
speech during interaction. Convergence occurs when speakers adjust their
normal speech to make it more similar to their interlocutor’s speech or to a
stereotype of it (i.e. the speaker converges towards some prestige norm that
they believe their interlocutor values). Divergence occurs when the opposite
takes place—speakers seek to make their speech dissimilar from that of their
addressee. Speech maintenance occurs when speakers do not make any
changes. However, Giles views this as a failure to converge (the expected type
of behaviour) and therefore considers it a subtype of divergence. Both conver¬
gence and divergence can take place upwards or downwards. Upward con¬
vergence takes place when speakers adjust their speech in the direction of the
speech norms of persons of higher social status. It is the most common type
because it is based on the universal human desire for approval. Downward
convergence involves adjustments in the direction of the speech norms of per¬
sons of lower social status. Downward divergence involves speakers empha¬
sizing the non-standard features in their repertoire, while upward divergence
involves emphasizing the standard features.
Accommodation can take place at any level of language use—in the choice
of language used in a bilingual situation, or in terms of volume, speed, pro¬
nunciation, choice of vocabulary, and grammatical structures, and also in
discourse features such as length of turn and choice of topic in monolingual
situations.
An example (based on Giles 1971) will make this clear. Let us imagine that
a shop assistant who normally speaks standard English is communicating
with a customer who speaks a non-standard variety. If the shop assistant ad¬
opts some of the non-prestige forms used by the customer, downward conver¬
gence takes place. If the customer adopts some of the prestige forms of the
shop assistant, upward convergence takes place. This is what we might norm¬
ally expect to happen. But let us now imagine that the customer is trying to re¬
turn some article she bought yesterday and the shop assistant is refusing to
refund her the money it cost. The shop assistant may gradually diverge up¬
wards by emphasizing the standardness of her language, while the customer
may diverge downwards by resorting to the ‘vernacular’ forms of her dialect.
Giles and his associates have carried out a number of studies to test the
claims of speech accommodation theory. A good example is Thakerar, Giles,
and Cheshire (1982). They studied conversations between British nurses of
unequal ranks and found that higher-status nurses tended to decrease their
speech rate and become less standard in their speech (for example, they used
the glottal stop /?/ in place of It/ more frequently), while the lower status
nurses increased their speech rate and used a higher proportion of standard
forms. The researchers argued that this study indicated that the nurses were
converging towards a stereotype of their interlocutors’ speech.
 Variability in learner language 129

Speech accommodation is motivated by the attitudes that speakers hold to¬

wards their audience. The variation occurs on a moment-by-moment basis
during an interaction. It constitutes a kind of style-shifting that is audience-
designed. According to Bell (1984), speakers respond to three aspects of their
addressees in deciding what adjustments to make: (1) their personal charac¬
teristics, (2) the general style of their speech, and (3) their use of specific lin¬
guistic features. Bell also suggests that the pressures on speakers to
accommodate increase when speakers address large audiences and take part
in service encounters (for example, interactions in shops and restaurants).
One of the strengths of speech accommodation theory is that it recognizes
the central importance of the addressee in accounting for variation. It can also
account for the shifting patterns of variation which occur within a single in¬
teraction—more easily, perhaps, than is possible in the Labovian or dynamic
paradigms. Speakers can switch from convergence to divergence as they reas¬
sess their addressee during the course of an interaction. One disadvantage of
the model, claimed by Bell (1984), is that it rests on the concept of ‘response
matching’, and therefore provides only a responsive dimension. Bell argues
that stylistic variation also needs to take into consideration an initiative di¬
mension (i.e. where stylistic choice is used dynamically to redefine an existing
situation). Finally, speech accommodation theory cannot account for all pat¬
terned variability. Beebe (1982: 22) recognizes that there is also ‘ex-
travergence’ (i.e. ‘variation which cannot or should not be described as
converging towards or diverging from an interlocutor’) and lists several kinds
including the interlanguage variation that results from learning, communica¬
tion or performance strategies, and systematic variation according to the lin¬
guistic environment.
Speech accommodation theory has been used in SLA research by Beebe
(1981), Beebe and Zuengler (1983), and Zuengler (1989) among others.

Sociolinguistic models: some final comments

The sociolinguistic models belonging to the Labovian, dynamic, and social
psychological paradigms, which have been discussed in this section, offer the
L2 acquisition researcher a number of tools for investigating learner lan¬
guage. Each paradigm has contributed in different ways. The Labovian para¬
digm has provided an excellent means for examining stylistic variation in
learner language. The dynamic paradigm has served as a basis for exploring
how the learner’s interlanguage develops in terms of shifting form-function
relationships. Speech accommodation theory has contributed to the study of
how learners vary in the way they use their L2 repertoire according to ad¬
dressee factors operating in a particular interactional context.
There is, however, one important caveat regarding the use of sociolinguis¬
tic models in the study of L2 acquisition, in particular the Labovian and dy¬
namic paradigms. Sociolinguists like Labov, Bailey, and Bickerton are
130 The description of learner language

concerned with variability in the speech of social groups. Labov, for instance,
sees the regular patterns of variability that he found in the speech of New
Yorkers or BEV as part of their ‘langue’. This concept, which comes from the
work of de Saussure, is a social one. It recognizes that the values that comprise
a particular language may not be reflected exactly in the speech of an indi¬
vidual speaker. Bickerton (1975) also sees variability as a group phenom¬
enon, arguing that most of the individual Guyanese Creole speakers he
studied had relatively non-variable systems. As Preston (1989) has pointed
out, such a view of language does not fit easily with the goal of SLA research,
which is to describe and explain the acquisition of L2s by individual learners.
Nor is it clear whether the notion of ‘social group’ is applicable to many lan¬
guage learners. For example, it does not seem appropriate to suggest that the
L2 learners in a classroom in Spain or Japan make up a social group in the
same way as street gangs in New York City do. Thus, while it may be possible
to transfer the techniques used by sociolinguists such as Labov and Bickerton
to the study of individual learners, there are obvious problems regarding the
applicability of the ‘social’ explanations that sociolinguists have provided.

Psycholinguistic models

Whereas the models discussed above come from the fields of sociolinguistics
and social psychology, the ones we will consider in this section are psycho-
linguistic in nature. They concern the way in which speech is planned and
monitored by speakers.

Planning models

The concept of planning has been widely used in models of language produc¬
tion. As Crookes (1991: 115) points out, most models subdivide planning
into macro- and micro-planning. The former ‘concerns the long range se-
mahtic and syntactic organization of a sizeable chunk of speech’, whereas the
latter ‘is concerned with purely local functions, like marking clause boundar¬
ies and selecting words’ (Butterworth 1980b: 159, cited in Crookes 1991).
If, as de Bot argues, ‘many aspects of speaking are the same for monolin¬
gual and bilingual speakers’ (1992: 2), it should be possible to make use of a
general model of speech production, such as that proposed by Levelt
(1989)—the model de Bot favours. Such a model suggests a number of pos¬
sible psycholinguistic sources of variability, in accordance with different
stages of speech production:

1 In the ‘conceptualizer’, where decisions are taken regarding which vari¬

ety of language to use, in accordance with situational factors, and also
which specific communicative intentions are to be realized in the spoken
message.
 Variability in learner language 131

2 In the ‘formulator’, where the ‘pre-verbal message’ provided by the con-

ceptualizer is converted into a speech plan by selecting the appropriate
words from the lexicon and by applying grammatical and phonological
rules.
3 In the ‘articulator’, which converts the speech plan into actual speech.
4 In the ‘speech comprehension system’, which provides the speaker with
feedback regarding the presence of possible mistakes in the phonetic plan
or in overt speech and which also enables the speaker to make adjustments
in the ‘conceptualizer’.

De Bot emphasizes that in the Levelt model ‘the different components are at
work simultaneously’ (1992: 6) and ‘that various parts of the same sentence
will be at different processing stages’.
L2 variability research has focused somewhat narrowly on the effect of
‘planning time’ on this production process, influenced no doubt by the fact
that, whereas LI production is largely automatic, L2 production is often not,
so that the amount of time a learner has to plan the different processing stages
is likely to affect output. Ochs (1979), again discussing production in the LI,
distinguishes planned and unplanned discourse. The former is ‘discourse that
lacks forethought and organizational preparation’, while the latter is ‘dis¬
course that has been thought out and organized prior to its expression’ (1979:
55). The distinction constitutes a continuum, and Ochs emphasizes that most
of the discourse encountered in day-to-day communication falls at neither
end but somewhere in the middle. Ochs also notes that some types of dis¬
course are more plannable than others. Thus conversations are usually not
planned ahead, but more ritualistic speech events (such as sermons) are.
There are linguistic differences in the two types of discourse. In unplanned
discourse, Ochs found that speakers rely more on the immediate context to
help them convey their message, make use of syntactic structures that tend to
emerge early during acquisition (for example, demonstrative modifiers, ac¬
tive voice and present rather than past tenses) and make extensive use of repe¬
tition and word replacement. Danielewicz (1984) found that there were clear
differences in the unplanned spoken language taken from dinner table con¬
versations and the planned spoken language produced by the same speakers
in lectures. The differences were both global in nature (for example, the way
evidence was used to construct an argument) and specific (for example, in the
complexity of clause and sentence construction and the use of attributive ad¬
jectives and participles). Danielewicz’s study is important because it shows
that the distinction between planned and unplanned discourse is independent
of that between speech and writing.

Monitoring
Speakers may monitor their output (i.e. pay conscious attention to specific
elements of the utterance in order to correct or improve them). Morrison and
132 The description of learner language

Low (1983) make use of a similar production model to that of Levelt to distin¬
guish post-articulatory monitoring, which operates on overt speech, and pre-
articulatory monitoring, which occurs prior to the implementation of the
phonetic plan. Levelt (1983), cited in Crookes (1991: 116), similarly distin¬
guishes a ‘production theory of monitoring’, according to which learners re¬
spond to ‘alarm signals’ during the course of implementing a plan and make
appropriate adjustments, and a ‘perceptual theory of monitoring’, according
to which users compare the final result of the production process with their
original intention. Macro- and micro-monitoring can also be distinguished.
The former involves adjustments to the communicative goal of discourse and
sentence plans. The latter takes place as the speaker/writer begins the process
of filling out constituent plans with linguistic forms and involves the substitu¬
tion of one selected form with another, preferred form. Micro-monitoring
can be carried out on lexis, syntax, morphology, and the phonetic realization
of the utterance.

Psycholinguistic models: a final comment

The conditions under which speech has to be produced will vary, and lead to
differences in the actual language that is used. A speech production model
such as Levelt’s enables us to recognize that whereas some variability is soci¬
ally motivated (for example, in message generation in the ‘conceptualizer’),
some (such as the construction of a phonetic plan in the ‘formulator’) is not.
Instead, it reflects the tension between the amount of effort needed to access
words from the lexicon, execute a grammatical encoding, and then assign a
phonological coding, and the availability of planning time. Some items in the
lexicon are likely to be more readily accessible than others. Production in¬
volves a constant trade-off of the competing demands on memory and control
mechanisms. The result will be systematic differences in language use in ac¬
cordance with opportunities available for planning.
Like sociolinguistic and social psychological models, psycholinguistic
models of L2 variability have their limitations. The psycholinguistic aspects
of language production that figure in L2 variability research draw heavily on
the notion of ‘attention’—with regard to both what is attended to, and the ex¬
tent to which the attention is conscious or not. In this respect, they do not dif-
fer from the Labovian paradigm, which, as we saw, treats attention as the
psycholinguistic mechanism through which social and stylistic factors affect
performance, and are subject to the same criticisms as this construct. Also, the
concepts of planning time and monitoring cannot easily account for the kind
of within-conversation shifting identified in speech accommodation models
and, in particular, for the influence the addressee exerts on performance.
Clearly, the study of L2 variability requires both a sociolinguistic and a psy-
cholinguistic perspective.
 Variability in learner language 133

The effects of planning time on L2 use have been investigated by myself

(Ellis 1987b) and Crookes (1989). Monitoring plays a key role in Krashen’s
Monitor Model (see Krashen 1981) and in a study by Hulstijn and Hulstijn
(1984).

Type of model Main constructs Methods Key SLA studies

Sociolinguistic:
Labovian paradigm Inter- and intra¬ Tasks designed to Dickerson 1975;
learner variation; elicit varying Beebe 1980; Tarone
style shifting occurs attention to speech; 1985; Young 1988
as a result of varying variable rules;
attention to speech. VARBRUL statistical
models.
Dynamic paradigm Language change as Implicational scaling; Huebner1983 and
a source of form-function 1985; Ellis 1985c
variability; analysis.
environmental
weight; lectal
variation; code¬
switching; grammar-
internal variation;
free variation.
Social psychological Speech shifts Experimental Beebe 1981; Beebe
paradigm involving investigation of and Zuengler 1983
convergence and factors that induce
divergence; convergence and
importance of divergence.
addressee and of
speaker’s attitudes
to addressee’s social
group.

Psycholinguistic:
Speech planning Planned and Experimental Eliis 1987b; Crookes
models unplanned manipulation of 1989
discourse; demands discourse planning
of short term conditions.
memory.
Speech monitoring Macro- and micro- Experimental Hulstijn and Hulstijn
models monitoring; pre- and manipulation of 1984
post articulation conditions that affect
monitoring; monitoring.
production-based
and perception-
based monitoring.

Table 4.2: Summary of different types of models used to explain variability in

language production
134 The description of learner language

A typology of variability in learner language

This section draws on many of the constructs that have already been intro¬
duced (and which are summarized in Table 4.2) to present a typology of the
different kinds of formal variation that can be found in the use of natural lan¬
guages—including learner language. The outline of the typology is shown in
Figure 4.1. The typology addresses variation in choice of linguistic form. It
excludes functional variation (i.e. variation in the choice of language
function), which is considered in the next chapter.

linguistic
context
form-
systematic situational
function
variation context
variation
psycholinguistic
intra-learner _context
variation

horizontal
non-systematic
— variation —
(free) variation
(synchronic)

variation
in linguistic— inter-learner
form — variation

vertical
— variation
(diachronic)

Figure 4.1: A typology of variation in the choice of linguistic form found in learner
language

A basic distinction is made between horizontal and vertical variation. Hori¬

zontal variation refers to the variation evident in learner language at any
single time, while vertical variation refers to variation over time and is,
therefore, coterminous with ‘order/sequence of development’. There is some
evidence to suggest that horizontal variation mirrors vertical variation for, as
Widdowson puts it, ‘change is only the temporal consequence of current vari¬
ation’ (1979a: 195), a view that accords with the role of change in the
dynamic paradigm.
Horizontal variation can be further subdivided into inter- and intra-learner
variation. Inter-learner variability reflects individual learner factors such as
motivation and personality, but it also arises as a result of social factors such
as social class and ethnic grouping, as the Labovian paradigm has demon¬
strated. As these same social factors are also involved in intra-learner vari¬
ation, there is clearly an interaction between individual learner factors such as
 Variability in learner language 135

sex and social class and the situational factors involved in style-shifting; as we
noted earlier, the markers involved in stylistic variation also function as social
indicators. In this chapter the focus is on intra-learner variation; Chapters 6
and 8 examine the social and psychological sources of inter-learner variation.
Intra-learner variation can be both systematic and non-systematic (altern¬
atively called free variation). 1 his distinction is, however, somewhat contro¬
versial, as many sociolinguists consider that free variation does not exist or
that it occurs for only a very short period of time and is of minor interest.4
Gatbonton (1978) and I (Ellis 1985c; 1989c) have, however, argued that free
variation constitutes an important mechanism of development.
Systematic variability arises as a result of external factors to do with the lin¬
guistic context, the situational context, and the psycholinguistic context. The
linguistic context concerns the elements that precede and follow the variable
structure in question. The situational context covers a whole host of factors.
Brown and Fraser (1979) classify ‘situation’ into ‘scene’ and ‘participants’.
Scene involves ‘setting’ and ‘purpose’, which can in turn be further subdiv¬
ided. Participants can be considered in terms of their individual or social char¬
acteristics and in terms of the relationships that exist between them. Preston
(1989) also offers a detailed breakdown of situational factors (for example,
time, topic, purpose, and tone). It is probably true to say that to date SLA
research has examined only a few of these situational factors. The psycho-
linguistic context refers to the extent to which the type of language use affords
time for planning and encourages or discourages monitoring.
Form-function variation is also systematic and is contextually induced. All
three types of context—linguistic, situational, and psycholinguistic—may in¬
fluence which forms learners employ to perform specific language functions.
Form-function variation is identified by means of a form-function analysis.
Learners construct networks of form-function relationships, and the descrip¬
tion of these networks provides clues about the nature of the learner’s internal
system.
Finally, we can also talk of task-induced variation. This is best considered
as a blanket term to cover the variability evident when learners perform
different tasks, and is ultimately traceable to one of the other sources (the lin¬
guistic, situational, or psycholinguistic context). We include it here because in
some studies no attempt has been made to identify the specific sources of vari¬
ability induced by different tasks.
Young (1989) suggests that the description of variability in learner lan¬
guage involves two processes. The first is to describe how changes in one
sociolinguistic variable influence variation in form. The second is to evaluate
the relative impact on variation of different sociolinguistic variables. How¬
ever, as Young points out, relatively few studies have addressed the second
question, so we will be concerned primarily with the first question.
The review of the research that now follows begins by looking at free vari¬
ation. This leads into a discussion of form-function variation. A detailed
136 The description of learner language

examination of systematic variation in relation to context follows. This starts

with an account of studies which have reported general task-induced
differences in learner output, and then moves on to examine form-function
relationships and the effects of specific contextual variables on learner out¬
put. Finally, a multi-factor approach is outlined.

Free variation in learner language

In Ellis 1985c, I reasoned that free variation can be considered to occur when
two or more forms occur randomly in (1) the same situational context, (2) the
same linguistic context, (3) the same discourse context, (4) perform the same
language function, and (5) are performed in tasks with the same processing
constraints. In other words, in order to claim that forms are in free variation,
it is necessary to demonstrate that there is no feature of the linguistic, situ¬
ational, or psycholinguistic context that exerts any probabilistic pull on the
use of one form in preference to another.
As an example, I referred to data from my study of three classroom learn¬
ers. One of the learners, ‘J’—a 10-year-old Portuguese boy—produced the
following two utterances in close proximity to each other during a game of
word bingo:

No look my card.
Don’t look my card.

I noted that the ‘don’t’ negative was the single instance out of 18 spontaneous
negative utterances produced during the first month of the study and was
used in an identical way to the predominant ‘no’ + V form of this period. A
further example from the same data can be found in one of the other learners’
use of copula ‘be’. ‘R’—a 10-year-old Punjabi-speaking boy—used the zero
and contracted copula in successive utterances, produced while pointing at a
map:

' There church.

There’s church.

I was able to detect regularities in the overall use of copula according to lin¬
guistic context (see the section on page 143 in this chapter), but specific in¬
stances of apparent free variation such as the one above were conspicuous in
the data.
A number of longitudinal studies testify to apparent free variation in the
use of two or more forms in learner language. Cancino, Rosansky, and
Schumann (1978) found that their subjects made use of a variety of forms to
express negation at each stage of their development. Jorge (one of the five
learners they investigated), for instance, used two forms initially (‘no’ + V and
‘don’t’ + V). From the second month of the study, Jorge began to use four
forms (the previous two plus an auxiliary negative and analysed ‘don’t’), and
 Variability in learner language 137

continued to do so for the remaining eight months of the study. Cancino et al.
were unable to write ‘rules’ to account for the use of the different forms, sug¬
gesting that they were in free variation.
Another possible area in which free variation appears to occur is in the use
of English simple and progressive verb forms in early learner language.
Wagner-Gough (1975), in a longitudinal study of Homer, a Persian boy, re¬
ports that he used both verb forms for an identical range of functions over an
extended period of time. Eisenstein, Bailey, and Madden (1982) also found
that their beginner learners used the two verb forms indiscriminately, while
their more advanced learners used them more systematically. Vogel and
Bahns (1989) report on the general pattern of acquisition of the two verb
forms, based on data collected by Wode in his longitudinal study of the natur¬
alistic acquisition of English by five German children (see Wode 1981). They
also found evidence of free variation. One verb after another appeared in its
inflected {-ing) form and its simple form before the learners distinguished the
use of the two forms functionally.
Free variation can also occur in learners’ use of pronoun forms. Nicholas
(1986) found that his subject, Cindy—a 3-year-old learner of L2 Ger¬
man—rapidly acquired three first person pronoun forms (ich, mich, and mir)
but used them variably to perform the same functions. Nicholas claims that
Cindy set about diversifying her L2 system by incorporating as many features
into it as possible. She recognized that the same function could be expressed
by a range of related forms and this motivated her to explore the potential of
variation.
A general finding of these studies is that free variation occurs during an
early stage of development and then disappears as learners develop better or¬
ganized L2 systems, a view of acquisition first put forward by Gatbonton
(1978).5 According to her gradual diffusion model, there are two broad stages
of L2 development: an ‘acquisition phase’ and a ‘replacement phase’. In the
former, the learner first uses one form in a variety of situations and contexts,
and then introduces another form which is used in free variation with the first
in all contexts. In the replacement phase, each form is restricted to its own
contexts through the gradual elimination of the other in first one context and
then another. Gatbonton developed this model to account for the patterns of
variation which she found in the production of three phonological features
(/0/, 15/, and /h/) in the speech of 27 French-Canadian learners of English eli¬
cited by means of reading-aloud and spontaneous-speaking tasks. In Ellis
1985c, I extended Gatbonton’s model to account for the elimination of free
variation as learners come to use each form to perform specific functions.
It is important to recognize that this account of free variation refers to the
early stage of acquisition, when a feature first appears in a learner’s language.
It can be tested effectively only by longitudinal studies that provide informa¬
tion about what learners do when they first acquire a new feature. It is not
possible to argue, as has Young (1993: 86), that the apparent absence of free
138 The description of learner language

variation in data collected cross-sectionally from learners designated ‘low

level’ on a general proficiency test constitutes evidence against free variation,
as such data does not show us how learners handle the acquisition of specific
linguistic features from one time to the next. It is crucial to do this to test my
claims about free variation.
Nevertheless, care needs to be exercised in making claims about free vari¬
ation. As Tarone (1988: 113) and Preston (1989) have pointed out, it is al¬
ways possible that what appears to be free variation is, in fact, systematic and
that the researcher has simply failed to uncover the factors that make it sys¬
tematic. In the last analysis, it is not possible to know all the factors (in par¬
ticular the psychological ones) that influence language use. Also, even if it
does exist it may not be of much significance for understanding how learners
develop their L2 competence. Whereas Gatbonton and I see free variation as
an important step in the acquisition process, Schachter (1986a) considers it of
little interest on the grounds that it occurs before the onset of the productive
use of a new feature. She comments: ‘Those isolated occurrences of a struc¬
ture prior to onset remind me of the puts-puts of a motor just before it catches
on with a roar.’ For her, acquisition begins ‘where the structure/form occurs
across different lexical items with some (unquantified) regularity’ (1986a:
127).

Systematic variation
We begin by considering task-induced variability and form-function variabil¬
ity. In both cases, the variability is context-induced. A particular task creates
specific contexts of use which influence the forms a learner chooses to use.
Similarly, the choice of a particular form to realize a particular function is in¬
fluenced by linguistic, situational, and even psychological contexts. How¬
ever, the research which we will consider first has not sought to identify the
effect of specific contextual factors on L2 use and, therefore, is best con¬
sidered apart from that which has. The bulk of this section, however, will
consider research that has examined the effect of these specific factors.

Task-induced variation

A number of studies have made use of two or more instruments to collect

data. These studies testify to the general effect that context has on learner out¬
put but they do not tell us anything about the specific causes of variability.
Tarone (1988: 61) provides a comprehensive list of such studies, compiled to
refute the claim by Swan (1987) that variation in learner language was less
common than she (and others) have claimed.
Two such studies have been mentioned in previous chapters. Larsen-
Freeman (1976b) used five different tasks—speaking, listening, reading,
writing, and elicited imitation—in her study of grammatical morphemes. She
found different accuracy orders, in particular between those tasks involving
 Variability in learner language

speaking of some kind and the reading/writing tasks (see Chapter 3, page 91).
Lococo s (1976) error analysis study also found that the number of errors
made by adult elementary learners of L2 Spanish varied according to task.
For example, preposition errors were more common and adjective and deter¬
miner errors less common in the translation task than in the free composition
and picture description tasks. Lococo suggested that the learners’ perception
of the task might be one factor influencing the results. Thus, the learners
might have focused on accuracy in the translation task and on expressing
ideas clearly in the picture description task.
The results of several studies lend support to the claim that differences in
learner language will occur according to whether the task requires something
approximating to normal communicative language use, as, for example, in a
recorded conversation, or less conventional language behaviour, as, for ex¬
ample, that elicited by a grammaticality judgement task. Schmidt (1980) in¬
vestigated second-verb ellipsis in sentences like ‘Mary is eating an apple and
Sue a pear’, and found that learners from a variety of language backgrounds
always included the second verb in free oral production, but increasingly
omitted it in proportion to the degree of monitoring that different tasks (im¬
itation, written sentence-combining, and grammaticality judgements) were
hypothesized to permit. Bahns and Wode (1980) reported differences in two
German children’s use of English negative constructions in naturally oc¬
curring speech and structured interviews (which involved translation). They
found evidence of clear form-function distributions for ‘don’t’ and ‘didn’t’ in
the spontaneous data but not in the elicited data. Finally, Hyltenstam (1984)
investigated pronoun copies in relative clauses and negatives in the L2 Swed¬
ish of learners from different language backgrounds (see Chapter 3 for an ac¬
count of pronoun copies). Hyltenstam found that some tasks (for example,
written composition) failed to elicit the use of these structures, other tasks
(for example, written grammaticality judgements) produced data that ‘lacked
patterning’, while yet other tasks (for example, an oral picture description
task) produced patterned data that showed a definite acquisition sequence.
Hyltenstam speculated that different tasks resulted in both quantitative dif¬
ferences (i.e. more or less target-like behaviour) and qualitative differences
(i.e. different, even contradicting rules).
Task-induced variability of the kind reported in these studies makes it diffi¬
cult to decide whether a learner can be said to ‘know’ or ‘not know’ a particu¬
lar linguistic form. It raises the whole question of what is meant by
‘knowledge’ of an L2—a question we will address in the concluding section
of this chapter. It also makes it difficult to plot developmental patterns. Some
researchers interested in linguistic universals have preferred to work with
data derived from a single task, often a grammaticality judgement task, to
avoid these problems. Chaudron (1983a), in a review of this kind of task, has
argued that the data it elicits mirror those obtained from natural language
use. However, the studies considered above, and others reviewed below,
140 The description of learner language

suggest that this claim is not justified. Substantial differences arise depending
on whether the task permits or encourages the use of metalingual knowledge,
or whether it taps ‘communicative’ behaviour. It is precisely this kind of vari¬
ation which researchers such as Tarone have set out to explore and to account
for.

Form-function relationships in learner language

Tarone (1988) suggests that the study of the way learners use particular forms
to express particular functions is relevant to the study of interlanguage vari¬
ation because ‘it provides a method of analysis which can reveal the linguistic
system hidden in a learner’s apparently unsystematic use’ (1988: 54). A good
example of this can be found in the study by Schachter (1986a), already
briefly referred to.
Schachter set out to re-examine the data for one of Cancino, Rosansky, and
Schuman’s learners (Jorge) in order to establish whether his use of negative
forms was as random as the original authors suggested. She argued that the
variability was not explicable in terms of different situation requirements, as
the data collection took place on a regular basis with the same situational
constraints throughout. However, Schachter was able to find evidence of ‘a
rich system, complex from the very beginning, which became even more so as
time progressed’ (1986a: 123-4). The main basis for this claim was a form-
functional analysis. Schachter identified seven functions performed by Jorge’s
productive negative utterances and found surprising regularity in his pairing
of forms and functions. For example, the formula ‘I don’t know’ was always
used to perform the same function of ‘no information’ (i.e. to indicate that the
speaker is not in a position to confirm or deny whether something is the
case), while ‘no + V’ with one exception carried the ‘denial’ function (i.e. to
assert that an actual, supposed, or proposed state of affairs does not hold for
the speaker). Only ‘no’ by itself was functionally ubiquitous. Schachter’s
study is important because it shows how essential it is to investigate form-
function relationships in learner language in order to understand the kind of
internal systems that learners build.
Even more powerful evidence of systematicity in the way a learner uses lan¬
guage to realize different functions is to be found in Huebner’s (1979; 1983)
longitudinal study of Ge, a Hmong learner of English. Huebner was able to
show that although Ge did not use forms to perform the same set of functions
that they performed in the target language, he did nevertheless establish sys¬
tematic form-function relationships. Huebner began by identifying a number
of linguistic forms for analysis, one of which was Ge’s use of articles (‘da’ and
zero article). He analysed these in terms of two binary categories of semantic
function:
+/- information assumed to be known by the hearer (HK)
+/- specific referent (SR)
 Variability in learner language 141

which, when combined, yield four categories of noun phrases, as shown in

Table 4.3.

Type Standard English forms Examples

1 -SR/+HK ‘the’, ‘a’, or zero Lions are beautiful.

2 + SR/+HK ‘the’ Ask the man over there.
3 +SR/-HK ‘a’ or zero She gave me a present.
4 - SR/- HK ‘a’ or zero He’s a nice man.

Table 4.3: Noun phrase types

Initially, Ge used ‘da’ mainly for (2), unless the noun phrase in question
functioned as a topic of the sentence, in which case he used zero article. Later
he used ‘da’ for all four types of noun phrase. Later still, he dropped ‘da’ for
type (4) and finally for type (3). In a subsequent paper, Huebner (1985) also
shows how Ge introduced ‘a’ with type (3) noun phrases. Huebner’s main
point is that what appears to be random use of articles in an obligatory occa¬
sion analysis, turns out to be highly systematic in a form-function analysis.
The same functional analysis of the use of articles was also employed in an¬
other study (Tarone and Parrish 1988). They reanalysed data which Tarone
(1985) had collected using three different tasks: (1) a grammaticality judge¬
ment test, (2) an oral interview, and (3) an oral narration task. Tarone and
Parrish re-examined the data from the two oral tasks. The accuracy level of
the (ten Japanese and ten Arabic-speaking) learners’ use of articles varied ac¬
cording to task. Tarone and Parrish were able to show that the different tasks
elicited different types of noun phrases to a different extent, and that this
went a long way to explaining the overall pattern of task variation. For ex¬
ample, both accuracy and use of Type 2 articles was greater on the narrative
than on the interview task. It was this combination that accounted for the
overall difference in performance on the two tasks. The point here is that the
different tasks favoured the use of different noun phrase types because they
required the performance of different functions involving articles. Again, this
study demonstrates that the communicative function of particular forms
must be taken into account if we are to understand the underlying system-
aticity of learner language.
There have been several other studies that have investigated form-function
relationships in learner language. Pfaff (1987a) studied oral narratives pro¬
duced by seventh grade Turkish learners of L2 German and found clear evid¬
ence of form-function mapping. For example, the learners referred to
protagonists in their narratives using pronouns, but used article + noun to re¬
fer to other participants. Several studies (for example, Godfrey 1980; Vero-
nique 1987) show that learners vary their use of syntactic devices according
to whether information is foregrounded or backgrounded. Williams (1989)
142 The description of learner language

found that standard English speakers, speakers of Singaporean English, and

L2 learners from a variety of language backgrounds used subject pronouns in
very similar ways. They were likely to drop them if the identity of the subject
was clear from the ongoing discourse and to use them when it was not. The
general conclusion to be drawn from these studies is that learners will express
what is important to them, sometimes realizing a function linguistically and
sometimes leaving it to be inferred with the help of various non-linguistic
clues.
The following are the main conclusions that can be drawn from the re¬
search to date:

1 A form-function analysis can identify systematic patterns of variability

where a target-based analysis has failed.
2 Learners construct form-function systems in the process of learning and
using an L2. These systems are likely to differ from the form-function
systems found in the target language.
3 The learner’s form-function systems evolve over time. Thus at any stage
of development different form-function systems are likely to be observed.
4 Learners will seek to use their linguistic resources to perform those func¬
tions that are communicatively important for accomplishing specific
tasks.

Form-function analyses of learner language are very promising because they

seem to afford a window through which to view how learners construct their
L2 systems. One problem, as Tarone (1988) has observed, is that the term
‘function’ has been interpreted very loosely. It includes ‘pragmatic function’
(for example, requests and denials), ‘discourse function’ (for example, topic
and cohesion), ‘semantic function’ (for example, specific and non-specific in¬
formation), and even ‘grammatical function’ (for example, subject and ob¬
ject). Clearly these are very different aspects of language. Ultimately, it will be
necessary to show how learners use the forms they have acquired to perform
all these functions, and also how the different kinds of function interact in de¬
termining linguistic choice—a potentially huge undertaking.

Context-induced variation

The studies of task-induced variability and of form-function relationships

discussed in the previous section demonstrate that much of the variability in
learner language is systematic. Learners make use of their L2 resources in ac¬
cordance with context. We now turn to research which has examined the ef¬
fects of different aspects of context: linguistic, sociolinguistic, and
psycholinguistic.
 Variability in learner language 143

The effects of linguistic context

Irrespective of the kinds of data used to study learner language, variability in¬
duced by the linguistic context occurs. This kind of variability, therefore,
does not appear to be strongly task-induced. The effects of the linguistic en¬
vironment are evident at the phonological, morphological, and syntactic
levels of language.
One of the earliest studies to investigate the effects of linguistic context on
learners’ production was Dickerson’s (1975) study of ten Japanese learners’
production of English /z/. Dickerson examined four different phonological
environments and found that in a dialogue reading task, the learners used the
correct target-language form whenever /z/ was followed by a vowel, but were
progressively less accurate in the other environments (+ consonants such as
/mI and /b/, + silence, and + consonants such as /9/ and /t/). Dickerson col¬
lected data at three different points in time and found that the same pattern
occurred at each. Although the learners improved in their ability to use target-
language /z/ in the more difficult contexts over time, the environmental effects
observed at time 1 were also evident at time 3.
The form of a word can affect the learners’ use of specific grammatical fea¬
tures. Thus, Wolfram (1989) reports that Vietnamese learners of L2 English
in the United States were more likely to manifest past tense marking on sup¬
pletive forms (for example, ‘go/went’) than on replacive forms (for example,
‘make/made’). Wolfram suggests that this may reflect a ‘principle of percep¬
tual saliency’, according to which the more distant the past tense form is
phonetically from the present tense form, the more likely it is to be more
marked for past tense. Whereas Wolfram’s study demonstrates an effect for
the phonetic form of a word, Saunders’ (1987) study indicates that the gram¬
matical form is also influential. He studied Japanese learners’ production of
the -s morpheme in consonant clusters (-ps, -ts, and -ks) in nouns and verbs.
He found that they were more likely to omit or shorten the -s in verbs than in
nouns. Saunders suggests that this may be due to the ‘amount of information
contained in the morpheme; the third person singular is almost always re¬
dundant while the plural usually gives some information not available in the
rest of the sentence’ (1987: 261).
Saunders’ explanation accords with what Young (1993) has referred to as
the functional hypothesis. This predicts tha't ‘the grammar of a language will
contain a constraint which prevents the deletion of elements when their
absence would result in loss of semantic information’ (page 78). Thus, a fea¬
ture like plural -s can be expected to be deleted in contexts where plurality is
marked by some other means (such as a quantifier such as ‘many’), but util¬
ized when there is no such marker. Young’s review of studies that have exam¬
ined this hypothesis shows very mixed results, while his own study of plural
marking in Chinese and Czech/Slovak learners of English, however,
produced little support for the hypothesis.
144 The description of learner language

Another issue of considerable interest is the relationship between linguistic

context and acquisition over time. One possibility is that learners begin with a
single feature which they use in all environments (i.e. initially learners display
categorical use of a language form), then manifest free variation as they ac¬
quire new variants for the form before progressing to a stage when they dis¬
tribute the variants systematically according to context. Dickerson’s study,
however, found no support for this. Systematic variation was evident at time
1 in her study, although it should be noted that her subjects were not complete
beginners. However, Gatbonton’s (1978) study of English /0/, /5/, and /h/ in
the speech of 27 French-Canadians does support a diffusion model (see the
discussion on page 137). Gatbonton obtained independent measures of the
subjects’ proficiency and found that in the least proficient (those in the
‘acquisition stage’) a single variant or free variation was evident, whereas in
the more proficient (those in the ‘replacement phase’) systematic variation in
accordance with linguistic context occurred. However, as Tarone (1988)
points out, Gatbonton’s study is cross-sectional in design and thus does not
constitute totally convincing evidence for acquisition as a process of gradual
diffusion.
Unfortunately, there have been few longitudinal studies that have investig¬
ated the influence of the linguistic environment. In Ellis 1988a, I investigated
the effects of linguistic context on two morphological features (3rd person -s
and copula -s) in the English speech of three classroom learners (one Portu¬
guese and two Punjabi-speaking) over a two year period. For purposes of
space, only results for copula -s (which is a variable form in native speaker
use) will be considered here. The main findings were:

1 Three variants of copula -s were identified (zero, full, and contracted).

2 These variants emerged in the same order for all three learners. The learn¬
ers first used the zero form, then the full form, and finally the contracted
copula.
3 The target-language variants (full and contracted copula) were used
more consistently when the preceding subject was a pronoun than when
the preceding subject was a noun. Conversely, the non-target-language
variant (zero copula) occurred more frequently when the subject was a
noun.
4 Some pronoun environments also favoured the use of the target-language
variants more than others.
5 The acquisition of the target-language variants took place initially in ‘easy’
environments (i.e. those involving a closed class of items), and then ap¬
peared to spread to more ‘difficult’ environments (i.e. those involving an
open class).
6 None of the learners achieved the pattern of variability for copula use
which has been reported for native-speaker speech (i.e. they did not acquire
the target-language variable rule).
 Variability in learner language 145

This study, therefore, suggests that the acquisition of target-language forms

may be closely linked to their use in specific linguistic environments. It pro¬
vides some support for the diffusion model.
Finally, linguistic context affects the choice of syntactical features. Hylten-
stam (1977) showed that two environmental factors had an effect on where
learners of L2 Swedish placed the negator; whether the negative clause was a
main or subordinate clause and whether the finite verb was an auxiliary or a
lexical verb. Using implicational scaling he showed that the learners went
through several stages of development, mastering the placement of the negat¬
ive in first one environment and then another. This study, then, also indicates
that the acquisition of syntactic forms proceeds from one context to
another.
There have been several other studies that have investigated linguistic con¬
text (for example, Carlisle 1991; Bailey 1989; Veronique 1987), providing
fairly conclusive evidence that it plays a major role in determining the process
by which phonological, morphological, and syntactic features are acquired.
A number of issues remain unclear, however. First, it is still not certain what
the learner’s starting point is. Is it categorical use of a single form followed by
free variation, as Gatbonton and I propose? Or do learners begin with the sys¬
tematic use of two or more variants in accordance with linguistic context, a
view Dickerson and Tarone favour? Second, there is some controversy as to
whether patterns of variability can best be explained in terms of form-
function relationships or in terms of linguistic context. For example, Kumpf
(1984) suggests that the variable use of past tense forms is governed by
whether the information is foregrounded or backgrounded, whereas Wolf¬
ram (1985) claims that the main factor is linguistic context. The position that
has been adopted in this chapter is that both factors exert an influence and, in
fact, may not easily be separated. Ideally, information regarding how the two
factors interact is desirable, but this is not yet available. Thirdly, it is not clear
to what extent the idea of systematic ‘diffusion’ is applicable to all learners.
Meisel, Clahsen, and Pienemann argue that ‘learners differ greatly with re¬
spect to which context is most suitable for the application of a new rule’
(1981: 126). But several studies (such as Carlisle’s study of epenthesis in ini¬
tial consonant clusters) have been able to show that at least some of the effects
of linguistic context hold good across learners.

The effects of situational context

The basis for many of the studies considered in this section was Labov’s
notion of style shifting. Tarone (1982) suggests that L2 learners possess a
continuum of styles, ranging from the superordinate (or careful) to the
vernacular. The former reflects the kind of language found in formal situ¬
ations that require careful language use, while the latter is evident in informal
situations that permit more spontaneous language use. Each style has its own
146 The description of learner language

linguistic norms. Learners style shift between the two styles in accordance
with the demands of the situation. Tarone argues that both the superordinate
and vernacular norms constitute part of the learners’ overall language com¬
petence or, as she later calls it, their capability (Tarone 1983). It follows that
in order to investigate this it is necessary to collect data that reflect different
sets of norms.
To study style shifting, Tarone adapts Labov’s methodology. She accepts
that the psycholinguistic mechanism responsible for style shifting is attention
to speech and suggests that different styles can be elicited using tasks that
require different degrees of attention to speech. Figure 4.2 demonstrates the
relationships between task, attention, and style that Tarone recognizes. Thus,
to examine learners’ vernacular style it is necessary to collect ‘unattended
speech data’, while to investigate their careful style, data from grammaticality
judgement tasks can be used. Various other kinds of task elicit intermediate
styles.

vernacular style 2 style 3 style 4 style n careful \

style style
(more (more TL- or
pidgin-like) NL-like) /

unattended attended
\ t /
various elicitation
1
grammaticality
speech speech tasks (e.g. imitation judgement
data data and sentence data
combining)

Figure 4.2: Style shifting in learner language (from Tarone 1983)

A number of studies have used this methodological framework for in¬

vestigating variability. For example, Dickerson (1975), in the study already
referred to, used three different tasks to elicit data and found that the use of
the target-language variant of/z/ was used least frequently in free speech, and
most consistently when her subjects were reading word lists aloud; perform¬
ance on a task that required learners to read dialogues aloud was intermedi¬
ate. The learners increased their use of the target-language variant over time,
but manifested the same order of accuracy in the three tasks on each occasion.
Dickerson also found that her subjects were more likely use those non-target-
language variants that were closest to the target-language variant in tasks that
allowed them to attend more closely to their speech.
In early discussions of the stylistic continuum (for example, Dickerson
1974; Tarone 1979), it was assumed that the careful style had more correct,
target-like variants than the vernacular style. However, learners do not al¬
ways manifest the target-language variant in careful language use. Sometimes
they transfer forms from their LI, particularly if these forms have prestige
value in their speech community. Beebe (1980) found that her subjects (adult
Thai learners of English) produced fewer instances of the target sound, /r/, in
 Variability in learner language 147

formal language use than in informal language use. This was because they
used a prestige Thai variant of /r/, which they associated with formal lan¬
guage use in their own language. Beebe’s study and similar studies by Schmidt
(1977) and Major (1987) provide evidence of the complex nature of style
shifting in learner language. They show that the careful style may not always
be the most target-like. To accommodate this, Tarone (1983) revised her view
of the careful style to allow for use of either a target-language or T1 variant. It
is this view that is reflected in Figure 4.2. This raises the important, but appar¬
ently unanswered question, as to precisely when learners choose an LI vari¬
ant and a target-language variant as the norm for the superordinate style.
A later study by Tarone (1985) also testifies to the complexity of style shift¬
ing. She investigated three grammatical morphemes (third person -s, the art¬
icle, plural -s) and one grammatical structure (direct object pronouns) using
data collected from three tasks. In the case of third person -s, the results dem¬
onstrated the expected pattern of style shifting. That is, the learners were
most accurate in the test and least accurate in the oral narrative. Flowever, the
pattern was reversed for the article and direct object pronouns—the learners
were most accurate in the task that had been designed to require the least at¬
tention to form. Tarone attempted to explain these results by suggesting that
the narrative task led the learners to attend to discourse cohesiveness to a
greater extent than the other two tasks, the article and object pronouns ser¬
ving as important markers of discourse cohesiveness. In other words, the
functional demands of the task imposed unexpected demands on the learner’s
attention. It was this that led Tarone and Parrish (1988) to investigate form-
function relations in the use of the article (see the discussion on page 141).
The non-occurrence of style shifting is demonstrated by Sato’s (1985)
study of a 12-year-old Vietnamese boy’s acquisition of English. Sato collected
data over a ten-month period using three different tasks (free conversation,
oral reading of a continuous text, and elicited imitation of words and short
phrases). She looked at target final consonants and consonant clusters. The
accuracy rankings of the first of these on the three tasks changed from one
time to the next. Greater consistency was evident for the second feature. Sato
concluded that not all variables yield the same pattern of variation.
Sato’s findings should not be considered inconsistent with those of Tarone
or, indeed, Labov. Both Tarone and Labov have made it clear that not all lin¬
guistic features function as markers and, therefore, not all will style-shift. In
the case of Labov’s work, it is possible to predict which features are likely to
style-shift on the basis of whether they display sensitivity to social factors
(function as indicators). However, social factors may not affect many L2
learners (for example, foreign language learners in a classroom setting). This
raises the important question as to why some L2 features style-shift and
others do not. Sato suggested that one factor may be linguistic difficulty. Fea¬
tures that are particularly difficult may not be subject to much variation for
148 The description of learner language

the simple reason that learners cannot alter their performance whatever the
task.
These studies—and others—indicate that style-shifting is influenced by a
number of different factors (for example, the learner’s LI, the learner’s stage
of development, and the difficulty of the target-language feature). Although
the studies were informed by a sociolinguistic model designed to account for
how situational factors affect language use, they have not shed much light on
the relationship between the situational context and learners’ use of the L2.
This is because the nature of the link between attention and social fac¬
tors—the primary causative variables—is particularly unclear in the case of
L2 learners. In fact, as we have already noted, L2 researchers have used the
stylistic continuum to address the psycholinguistic causes of variability rather
than the sociolinguistic ones. Tarone (1988), in fact, treats L2 models based
on Labov as exemplars of ‘psychological processing theories’. This is under¬
standable given the problematicity of applying the concept of ‘social group’
to many L2 learners, which we discussed earlier. In effect, then, Tarone has
chosen to ignore what is central to the Labovian model, namely social or¬
ganization, with the result that the difference between sociolinguistic and
psycholinguistic models has been blurred.
Other studies have taken place within the social-psychological perspective
of Accommodation Theory, and these have proven more illuminative of the
relationship between social/situational factors and L2 use in so far as they
have examined what social factors motivate the use of psycholinguistic fac¬
tors such as attention. Studies by Beebe (1977) and Beebe and Zuengler
(1983) demonstrate that learners are sensitive to their interlocutor. The sub¬
jects (adults in the first study and children in the second) were Chinese-Thai
bilinguals who were interviewed twice on the same topics, once by a Chinese
interviewer and once by a Thai. Both interviewers, however, were fluent in
Thai. The bilinguals adapted their speech to their interlocutor by using more
Thai phonological variants with the Thai interviewer and more Chinese vari¬
ants with the Chinese interviewer. Beebe and Zuengler (1983) also report a
similar study involving Puerto Rican learners in English in New York, who
were interviewed in English by a monolingual native English speaker, an Eng¬
lish-dominant Hispanic, and a Spanish-dominant Hispanic. They found that
the learners used more dependent clauses with the monolingual English
speaker than with the English-dominant Hispanic. They suggest that learners
have difficulty in identifying with members of their own ethnic group when
speaking in the L2 and, therefore, do not converge. This idea receives some
support from Takahashi’s (1989) study of the effects of Japanese and non-
Japanese listeners on the speech patterns of Japanese learners of L2 English.
She found that the learners became more hesitant and briefer when ad¬
dressing a listener with the same native language background, and also were
less prepared to negotiate any communication problem. They also reported
feeling more uncomfortable.6 Dowd, Zuengler, and Berkowitz (1990) review
 Variability in learner language 149

the L2 research that has investigated ‘social marking’ by learners and con¬
clude that it can occur even during the early stages of learning, despite the
learners’ obvious limitations in repertoire, and that learners seem to be aware
of specific linguistic features that ‘stereotype’ native speakers of a language.
Learners also seem to be aware of their own identity as learners and, at
times, to exploit it. Rampton (1987), in a study considered in more detail in
the next chapter, noted that intermediate learners of L2 English sometimes re¬
sorted to ‘me no’ constructions (for example, ‘me too clever’ and ‘me no do
it’), even though they clearly possessed the competence to produce target-like
utterances. He suggests that the learners used this primitive negative con¬
struction as a means of impression management—to tone down the force of
their speech acts by drawing attention to their insignificant status as learners.
Once again, then, we see the importance of a form-function analysis.
Another aspect of the situation which learners have been shown to be sens¬
itive to is topic. Selinker and Douglas (1985) found that a Polish learner of L2
English varied markedly on a number of linguistic features according to
whether the ‘discourse domain’ concerned an everyday topic (telling your
life-story) or a specialized, technical topic (critical path schedules). Zuengler
(1989) also refers to evidence that shows that the attitude learners have to¬
wards a topic—whether they see themselves as experts or non-experts on it
relative to their native speaker interlocutors—will affect their language beha¬
viour, making them more or less likely to interrupt, for example.
It is clear that learners, like native speakers, are influenced by situational
factors and, in particular, by their addressee. Learners exploit their linguistic
resources in order to behave in sociolinguistically appropriate ways. Some¬
times this results in more target-like behaviour and sometimes in less, de¬
pending on who their addressee is, the particular language functions they
wish to perform, the topic of the discourse, and their perceptions of their own
expertise on a subject. In general, however, the research to date has been dis¬
appointing in its failure to tease out the contributions that different situ¬
ational factors make to overall variation in learner language.

The effects of psycholinguistic context

We have already seen that learners seem to vary in their use of linguistic forms
depending on where their attention is focused. The studies of style-shifting
considered in the previous section were based on the assumption that learners
will pay more or less attention to form in different tasks and that this will be
reflected in systematic variability. Thus, although these studies afforded little
insight into the role of situational factors, they do testify to the importance of
language processing factors. Unfortunately, however, ‘attention to speech’
remains a vague concept and cannot be easily demonstrated independently of
language performance (see the criticisms of the Labovian paradigm earlier in
this chapter).
150 The description of learner language

Other studies have specifically set out to examine the effects of psycho-
linguistic processing factors. In one such study, Hulstijn and Hulstijn (1984)
investigated the effects of time pressure, focus of attention (i.e. whether on in¬
formation or on linguistic form), and metalingual knowledge on the accuracy
with which two Dutch word order rules (Inversion and Verb-end) were per¬
formed in a story-retelling task in L2 Dutch. These rules operate in very sim¬
ilar ways to the same rules in L2 German (see Chapter 3 for a description).
The results indicated that neither time pressure nor metalingual knowledge
by themselves had any effect, but that focusing attention on form increased
accuracy in both structures. It should be noted, however, that the time factor
did influence two other aspects of the learners’ performance, response dura¬
tion and speech rate.
Two other studies have examined the effects of planning time on L2 pro¬
duction. In Ellis (1987b), I asked 17 adult learners of English to perform three
narrative tasks. Task 1 consisted of a written composition for which one hour
was allowed; Task 2 was an oral reproduction of the same composition (with¬
out recourse to the written version); Task 3 consisted of a different composi¬
tion which the subjects were asked to relate orally without any advance
planning. I compared the accuracy with which the learners used three past
tense morphemes (regular -ed, irregular, and copula). In the case of the regu¬
lar past tense, a clear pattern emerged. The learners were most accurate in
Task 1 and least accurate in Task 3, with Task 2 intermediate. This study
suggests, contrary to Hulstijn and Hulstijn’s study, that the availability of
planning time systematically affects the accuracy with which at least some
target variants are performed.7
The contradiction in the results of these two studies is apparent rather than
real. Clearly, it is what learners do with planning time that is important. If
they use it to focus on form (as opposed to other aspects of the speech produc¬
tion process), the result is likely to be increased accuracy. However, if they use
it to plan and organize the informational content of their output, an increase
in accuracy is less likely. This suggests, as Hulstijn points out, that ‘it is neces¬
sary to distinguish between task and task requirement’ (1989a: 29), as in fact
his own study did by distinguishing task in terms of time-pressure and task re¬
quirement in terms of attention to form.8 It must also be acknowledged that
not all linguistic forms are affected by planning. Those forms that conform to
some regular underlying pattern (like past tense -ed) are more likely to re¬
spond than those that do not.
Further evidence of substantial effects on L2 production resulting from
planning time comes from Crookes (1989). Forty intermediate and advanced
Japanese learners of L2 English were asked to complete tasks involving a de¬
scription of how to construct a Lego model and an explanation for the siting
of a set of buildings on a map. Some of the tasks were performed under a ‘min¬
imal planning condition’ (no planning time allowed) and others under a
‘planning condition’ (ten minutes was provided to plan words, phrases, and
 Variability in learner language 151

ideas). The tasks performed under the planning condition resulted in more
complex language as measured, for instance, by the number of subordinate
clauses per utterance. However there were no statistically significant differ¬
ences in general measures of accuracy, although the trend was in the direction
of the planning condition. The article ‘the’ was used more accurately in the
planning condition, but other grammatical morphemes (plural -s and article
‘a ) were not. The results of this study, like my own, were mixed; the accuracy
of some linguistic features increased as a result of planning time while that of
others did not.
The psycholinguistic context—like the linguistic and situational con¬
text—influences learner production, accounting for some of the overall vari¬
ability. Planned discourse is likely to manifest more frequent use of
target-language variants and also greater overall complexity than unplanned
discourse. However, the availability of planning time does not guarantee in¬
creased accuracy, as production involves learners in an intricate series of in¬
terlocking acts of planning, which compete for their attention. Thus, for
example, the effort required to plan the propositional content or to produce
complex sentences may inhibit the learner from attending to specific linguistic
forms. What is becoming clear, however, is that L2 production is charac¬
terized by predictable patterns.

A multi-factor approach

In the preceding sections we have been able to identify a number of factors

that induce variability in learner language. However, because most of the stu¬
dies we considered focused on single factors (for example, linguistic context,
audience design, or form-function relationships), it has not been possible to
establish the differential contribution of particular factors, nor how they in¬
teract. Clearly, though, to account for the complexity and multidimensional¬
ity of learner language, such a multi-factor approach is needed. In this
respect, a study by Young (1988a; 1991) is of considerable importance.
Young investigated one linguistic variable—marking plural -s on English
nouns—in the speech of twelve Chinese learners, who were interviewed in
English twice, once by a native English speaker and the second time by a fel¬
low Chinese speaker. Data relating to four general factors were obtained: (1)
the context of the situation (in particular the extent to which each learner
converged with the interviewers in terms of general social factors such as eth¬
nicity, sex, education, and occupation), (2) the subjects’ proficiency in Eng¬
lish (whether ‘high’ or ‘low’), (3) the linguistic context (whether the plural
nouns were definite or animate, the syntactic function of the noun phrase, and
the phonological environment), and (4) redundancy in plural marking
(whether plural -s was omitted because plurality was indicated by some other
linguistic device such as a numeral or verb-subject agreement).
152 The description of learner language

Young reports that the learners marked 65 per cent of nouns correctly for
plural -s. Using the VARBRUL computerized statistical procedure, Young
was able to calculate the effect that each factor had on the learners’ use of plu¬
ral -s. Table 4.4 summarizes the main results. As predicted, Young found that
all four general factors accounted for the variability present in the data. One
of the most interesting findings was that different factors influenced the per¬
formance of low- and high-proficiency learners. Thus, for instance, the
phonological environment of -s had a significant impact on variation only
during the early stages of acquisition, while social convergence with an inter¬
locutor had a significant effect only during the later stages. Another interest¬
ing discovery was that the presence of some other marker of plurality (for
example, a numeral) seemed to trigger -s. One reason for this was that the
learners, particularly those of low proficiency, made frequent use of a closed
set of ‘measure expressions’ (for example, numeral + years, days, hours,

Factor Low proficiency High proficiency Combined

proficiency

1 Social - + +
convergence with
an interlocutor
2 Definiteness - - -

3 Animacy + + -
4 Position of noun + + +
within noun
phrase
5 Syntactic function + + +
of noun phrase
6 Phonological + +
context (preceding
segment)
7 Phonological +
context (following
segment)
8 -s marked + + +
irrespective of
existence of other
+ markers of
plurality in noun
phrase
9 Noun-verb — — -
concord

+ = factor accounts significantly for observed variability

- = factor does not account significantly for observed variability

Table 4.4: The contribution of different factors to the variable production of

plural-s (simplified from Young 1988a: 293)
 Variability in learner language 153

dollars). These expressions were formula-like; the nouns in them were more
or less invariably marked with -s.
Another study that employs a multi-factor approach, again using
VARBRUL, is Adamson and Regan (1991). This investigates the acquisition
of variable -ing, which occurs in a number of English grammatical structures
including tenses, participles, and nouns and has two phonetic forms, [in] and
[iq], the latter constituting the prestige target-language variant. The subjects
included both native speakers and Vietnamese and Cambodian learners of
English. Hypothesizing that the learners would begin with [irj] as a result of
transfer from their LI, Adamson and Regan investigated the factors that con¬
tributed to the appearance of the non-prestige [in], arguing that this would
signal the learners’ integration into the speech community. Data were coll¬
ected using interviews designed to control for shifts of formality. The results
showed that the learners’ sex, the opportunity to monitor production by at¬
tending to speech, and the grammatical category of -ing all had a significant
effect on the variant used. Thus, males favoured the non-standard [m] to a
greater extent than females and, when monitoring, were more likely to use it.
The authors suggest that this reflected the males’ desire to match male native-
speaker norms. The results for the different grammatical categories suggest
that the spread of [m] from one category to another was influenced partly by
its frequency in the input, but also by whether the category in question was
open or closed (for example, [in] appeared early in words like ‘something’
and ‘nothing’, both of which of are members of a closed grammatical class).
These two studies provide some of the most convincing evidence to date
that variability in learner language is systematic. They testify to its enormous
complexity but also show that it is possible to account for it.

Summary
The research reported in the previous sections has focused on variability in
linguistic form (i.e. the starting point has been a specific linguistic form or
forms) in the output of individual learners. In accordance with the overall
goal of this section of the book, the main aim has been to describe rather than
to explain the variability. For this reason, there has been no in-depth discus¬
sion of theories of L2 acquisition based on variability. These are considered in
Chapter 9. The main characteristics of learner language variability are sum¬
marized below.
First, somewhat controversially, a number of researchers have claimed to
have observed instances of free variation (i.e. the unsystematic use of two or
more forms). In most cases evidence for this kind of variability is only found
in the rich data available in longitudinal case studies of individual learners. It
appears to be most common in the early stages of development and may rap¬
idly disappear. It has been suggested that learners go through an ‘acquisition
phase’, involving free variation as new features are acquired, and then a
154 The description of learner language

‘replacement’ phase, during which variability becomes systematic as the

learner sorts out when to use the new features.
Form-function analyses have shown that what appears to unsystematic
may in fact be systematic. A number of studies—again mainly longit¬
udinal—indicate that learners construct form-function networks in which in¬
dividual forms are used to perform specific functions. These networks may
not be target-like, and they evolve over time. They provide some of the stron¬
gest evidence of the ‘creativity’ of the L2 learning process and indicate the im¬
portance of going beyond a target-language-based analysis of learner
language.
Third, as the form-function analyses show, much of the variability in learn¬
er language is highly systematic. This systematicity is context-induced. The
effects of context are evident globally as task-induced variation. Quantitative
and qualitative differences in learner language can occur according to
whether the data come from tasks designed to tap natural language use, or
tasks intended to elicit knowledge of specific linguistic forms.
Specific effects for the linguistic, situational, and psycholinguistic context
have also been identified. Linguistic context has been shown to have an effect
on phonological, morphological, and syntactic features. Some environments
favour target-language variants, while others favour interlanguage variants,
suggesting that learners find it easier to use the target-language form in par¬
ticular environments to begin with. As they progress, so they master the use of
the form in the more difficult environments. There may be considerable inter¬
learner differences in the way linguistic context affects acquisition, however.
Situational context, which has figured so strongly in variability studies in
general, has received rather scant attention in SLA research. A number of stu¬
dies have demonstrated that style-shifting occurs in learner performance, but,
as has now become clear, learners do not always use the target-language vari¬
ant in tasks requiring formal language use and interlanguage variants in tasks
requiring informal language use. The learners’ stage of development, their
LI, and the linguistic feature under study all influence whether and in what
direction style shifting occurs. More promising is the research that indicates
that learners are sensitive to both their audience and the discourse topic, but
to date the research in these areas has been limited.
Although the L2 studies undertaken within the Labovian paradigm shed
little light on how situational variables affect learner performance, they did
demonstrate conclusively that attention (a psycholinguistic construct) is an
important factor. Other studies undertaken within a more explicit psycho¬
linguistic framework also show that planning time affects both the accuracy
with which specific forms are performed in different tasks, and also the over¬
all complexity of the language used. The attention that learners pay to differ¬
ent aspects of language is influenced by various factors, one of which is clearly
planning time.
A multi-factor approach to identifying sources of variability has been de¬
scribed. This suggests that the patterns of variability in the use of a single
 Variability in learner language 155

linguistic variable (such as plural -s) can only be fully identified and
accounted for if a range of possible factors that affect learner language are
taken into account. This approach also suggests that different factors may be
important at different stages of the learner’s development.
The research to date suggests that horizontal and vertical variability are
inextricably intertwined. This is most clearly evident in the studies that have
investigated linguistic context, form-function relationships, and planning
variability. Initially learners begin with ‘simple’ systems, characterized by the
use of a given form in ‘heavy’ environments only (often in closed classes) and
in planned discourse, and also by unique form-function mappings. Over time
they learn to use the form in a target-like way by extending its use to increas¬
ingly ‘lighter’ contexts and to unplanned discourse and by reorganizing form-
function networks so that they correspond more closely to those of native
speakers.

Conclusion

In this chapter we have explored the nature of variability in learner language

and have seen that there is ample evidence to show that variability is system¬
atic, traceable to particular causes. It remains for us to consider whether vari¬
ability is a significant phenomenon—whether it constitutes, as Tarone has
argued, ‘a phenomenon that must be accounted for by any theory of second
language acquisition’ (1988: 142). Here we find conflicting opinions, de¬
pending on whether the approach is a linguistic one following the Chom¬
skyan tradition or a sociolinguistic/psycholinguistic one.
As we saw in the introduction to this chapter, researchers in the Chom¬
skyan tradition argue that the primary goal of SLA research is to build a the¬
ory of L2 competence, and that in this respect variability is of no interest. As
Gregg (1990) puts it:

The variabilist is committed to the unprincipled collection of an uncon¬

trolled mass of data, running the real risk that the object of study will
become as Roger Brown once put it, ‘cognitively ... repellent.’

Although Gregg acknowledges that variability is ‘a fascinating and puzzling

phenomenon’, he believes that variable features are the exception rather than
the rule in learner language. In this he is surely mistaken, as the studies re¬
ported in this chapter provide indisputable evidence of extensive variability in
learner language. Gregg argues that although the study of variability might
contribute to a theory of performance, it will tell us nothing about compet¬
ence. Variability studies, he argues, describe, but Gregg seeks to explain. For
Gregg, knowledge is categorical—we either know a rule or we do not. He is
unable to accept the notion of a fuzzy, probabilistic knowledge and is particu¬
larly dismissive of variable rules.
156 The description of learner language

On the other hand, researchers in the functionalist tradition see the study of
language in its social context as necessary for theory-building. In responding
to Gregg’s arguments, Tarone (1990) quotes Romaine (1984: 78-9), who
takes a very different view of what knowledge of a language involves:

Rule acquisition is not an all or nothing affair... There may be a number of

aspects of the internal workings of a rule, some of which may be acquired
before others. There are also social dimensions of a rule relating to its use.

Here knowledge itself is seen as containing variability. Tarone argues that a

theory of L2 acquisition must account for the learner’s capability—the actual
ability to use particular rules that a learner has—and not just knowledge
about what is grammatically correct. For Tarone it is this capability rather
than competence that underlies actual performance. In Ellis 1990c I make a
similar point in arguing that even if competence (narrowly defined) is itself
not variable, the learner’s proficiency is. I used this term with the meaning
assigned to it by Taylor (1988)—the ability to use knowledge in specific
contexts.91 argued that for many researchers—especially those interested in
educational issues—it is not the learners’ competence that is important but
their proficiency. Both Tarone and I also maintain that the study of the
systematicity evident in learner language variability can help explain how
learners organize their L2, although we both accept that work in the
Chomskyan tradition that Gregg favours can also contribute to theory
development.
This debate is an old one and is likely to continue into the future. It rests on
how competence is defined, in particular on whether it is to be viewed nar¬
rowly in terms of ‘knowledge’ or more broadly in terms of ‘ability to use
knowledge’, both of which are seen as distinct from ‘performance’.10 As
Widdowson has pointed out in a discussion of these distinctions, ‘as soon as
you talk about competence as ability, or what people can actually do with
language, you get into all kinds of difficulty’ (1989: 134) . This is because it
necessarily involves consideration of a multitude of interacting factors, as in¬
deed the discussion of variability in this chapter has demonstrated. Whereas
‘universalists’ see this as a reason for treating competence as ‘knowledge’ and
so avoiding the ‘difficulty’ that Widdowson talks about, variabilists are pre¬
pared to accept it in order to undertake what they see as the essential task—to
account for competence as ‘ability’.

Notes

1 While many linguists—including those who work with L2 data—continue

to make use of data based on speakers’ intuitions regarding what is gram¬
matical, many now recognize the need to work with other kinds of data, in¬
cluding those derived from natural language use.
 Variability in learner language 157

2 The example of a variable rule is taken from Fserch (1980) rather than
Labov because it is presented in a form that is easy to understand. Readers
who would like to see examples of Labov’s variable rules can do so
(Labov 1972: 229, 240).
3 Preston (1989: 20) comments: ‘the psycholinguistic possibility for a vari¬
able rule does not seem to be arcane’.
4 Fischer (1958) states the position adopted by many sociolinguists on free
variation as follows:
‘Free variation’ is of course a label, not an explanation. It does not tell
us where the variants come from nor why the speakers use them in dif¬
fering proportions, but is rather a way of excluding such questions
from the scope of immediate inquiry.
5 Beebe (1982) puts forward a similar view to that of Gatbonton. She ar¬
gues that deviations in interlanguage phonology are relatively few ini¬
tially but then increase markedly as more variants are acquired and
compete in the learner’s interlanguage system before diminishing in the
later stages.
6 The claim that the difficulty L2 learners experience communicating with
members of their own ethnic group in the L2 is arguable. It may be that
this difficulty simply reflects the fact that they find using the L2 with nat¬
ive speakers of their own LI bizarre.
7 Planning variability cannot account for all the variability evident in learn¬
ers’ use of the past tense, as Wolfram’s (1989) review of studies that have
investigated systematic variability in L2 tense marking demonstrates. In
particular, Wolfson draws attention to a ‘lexical variable’ in the early
stages of acquisition. That is, learners vary in their ability to mark for past
tense according to the particular verb they are using.
8 One of the main differences between the Hulstijn and Hulstijn (1984) and
Ellis (1987b) studies and the style-shifting studies of Dickerson (1975)
and Tarone (1985) is that the former clearly do distinguish ‘task’ and
‘task requirement’ whereas the latter do not. The style-shifting studies use
different tasks to elicit what they presume to be differences in attention.
The psycholinguistic studies use the same task and vary the conditions un¬
der which it is performed.
9 Widdowson (1983: 8) uses the term ‘capacity’ to refer to ‘the ability to
produce and understand utterances by using the resources of the gram¬
mar in association with features of context to make meaning’. It is clear
that ‘capability’, ‘capacity’, and ‘proficiency’ have much in common.
10 This distinction between competence as abstract knowledge and compet¬
ence as ability to use knowledge was first discussed in Flyrnes (1971).
Flymes took the view that the latter definition was to be preferred. How¬
ever, other theorists, including those working in the field of education,
have adopted different viewpoints. Canale and Swain (1980), for ex¬
ample, opt for the narrower definition. It should also be noted that this
158 The description of learner language

distinction is separate from and cuts across the other frequently discussed
distinction in the literature—that between ‘linguistic competence’ and
‘communicative competence’. Thus, although Canale and Swain prefer to
exclude ‘ability to use’ from their definition of ‘competence’, they argue in
favour of including sociolinguistic, discourse, and strategic knowledge
within its compass.

Further reading
It is probably useful to begin your further reading with some background
information about sociolinguistics.
D. Preston, Sociolinguistics and Second Language Acquisition (Basil Black-
well, 1989), provides a thorough introduction. A somewhat more readable,
shorter but less comprehensive account is available in
R. Fasold, ‘Variation theory and language learning’ in P. Trudgill (ed.), Ap¬
plied Sociolinguistics (Academic Press, 1984).
Variability in learner language has attracted considerable attention. There
are a number of full length books devoted to it. The following are collections
of papers:
R. Ellis (ed.), Second Language Acquisition in Context (Prentice Hall Interna¬
tional, 1987).
S. Gass et al. (eds.), Variation in Second Language Acquisition: Volumes 1
and2 (Multilingual Matters, 1989).
M. Eisenstein (ed.), The Dynamic Interlanguage: Empirical Studies in Second
Language Variation (Plenum Press, 1989).
A survey of much of the research together with a clear account of the main
theoretical positions can be found in:
E. Tarone, Variation in Interlanguage (Edward Arnold, 1988).
For readers less concerned with detailed research reports and more interes¬
ted in the general approaches to L2 variability, two excellent sources are:
E. Tarone, ‘On the variability of interlanguage systems’. Applied Linguistics
(1^83)4:143-163.
L. Beebe’s chapter in L. Beebe (ed.), Issues in Second Language Acquisition
(Newbury House, 1988).
Readers may also wish to read more about different views regarding ‘com¬
petence’. Two sources are useful here:
Applied Linguistics, 10: 2 (This issue is devoted to a reconsideration of ‘com¬
municative competence’.)
Applied Linguistics, 11:4 (This issue contains Gregg’s attack on ‘variabilists’
and responses by Tarone and myself.)
5 Pragmatic aspects of learner language

Introduction

The focus of enquiry in the preceding chapters has been the formal linguistic
properties of learners’ interlanguages. As was made clear in Chapter 1, main¬
stream second language acquisition research has been primarily concerned
with these aspects of second language (L2) learning. Increasingly, however,
researchers are giving attention to pragmatic aspects of learner language.
This has been motivated in part by the belief that a full understanding of how
formal properties are learnt will not be achieved without examining the way
in which these properties are used in actual communication (hence the form-
functional analyses discussed in the last chapter). It has also been motivated
by the belief that the study of learner language requires a consideration of
pragmatic aspects in their own right. According to this view the goal of SLA
research is to describe and explain not only learners’ linguistic competence,
but also their pragmatic competence. The growing interest in interlanguage
pragmatics reflects the enormous developments in the theoretical and empir¬
ical study of pragmatics over the last two decades (see Levinson 1983;
Coulthard 1985; Hatch 1992, for surveys of the field).
Pragmatics is the term used to refer to the field of study where linguistic fea¬
tures are considered in relation to users of the language (Levinson 1983).
When speakers perform utterances in context they accomplish two things: (1)
interactional acts and (2) speech acts. The former impose structure on the dis¬
course by ensuring that one utterance leads smoothly to another; they con¬
cern how speakers manage the process of exchanging turns, how they open
and close conversations, and how they sequence acts to ensure a coherent
conversation. Speech acts constitute attempts by language users to perform
specific actions, in particular interpersonal functions such as compliments,
apologies, requests or complaints.1 They are considered in greater detail in
the following section.
This chapter will not attempt to provide a comprehensive overview of the
work undertaken in interlanguage pragmatics to date. Instead, we will adopt
what Kasper and Dahl (1991) refer to as the ‘narrow sense’ of ‘interlanguage
pragmatics—the performance and acquisition of speech acts by L2 learners.2
The justification for this decision lies in the fact that it is this aspect of prag¬
matics which has received the greatest attention in SLA research and that, al¬
though there is some work on how L2 learners perform interactional acts,
160 The description of learner language

there has been relatively little on how they acquire the ability to do so. How¬
ever, we will give some attention to impression management—the way learn¬
ers make use of their L2 resources in interaction to create social meanings
favourable to themselves—as this relates closely to work on speech acts.
Also, we will concentrate on the pragmatic aspects of learners’ spoken rather
than written language.

Speech acts and illocutionary meaning

According to speech act theory (Austin 1962; Searle 1969) the performance
of a speech act involves the performance of three types of act: a locutionary
act (the conveyance of propositional meaning), an illocutionary act (the per¬
formance of a particular language function), and a perlocutionary act (the
achieving of some kind of effect on the addressee). Searle (1975) distin¬
guished ‘direct’ and ‘indirect’ speech acts. In a direct speech act, there is a
transparent relationship between form and function as when an imperative is
used to perform a request (for example, ‘Pass me the salt.’). In an indirect
speech act, the illocutionary force of the act is not derivable from the surface
structure, as when an interrogative form serves as a request (for example,
‘Can you pass me the salt?’).
The successful performance of an illocutionary act is achieved when the
speaker meets a number of conditions associated with that particular act.
Searle (1969) distinguishes three types of conditions: preparatory conditions,
sincerity conditions, and essential conditions. For example, the illocutionary
act of ‘ordering’ is successfully performed when both the speaker and hearer
recognize that the speaker is in a position of authority over the hearer (prepar¬
atory condition), that the speaker wants the ordered act to be done (sincerity
condition), and that the speaker intends the utterance as an attempt to get the
hearer to do the act (essential condition). If any one of these conditions is not
met or is challenged by the hearer, the act may not be successfully performed.
For this reason, they are seen as of primary importance in a theory of speech
acts.
Other considerations of a secondary nature also enter into speech act per¬
formance—in particular, politeness (see Leech 1983 and Kasper 1990).
Speakers have to take account of their relationship with the addressee and the
degree of imposition on the addressee in order to ensure that harmonious so¬
cial relations between the speakers are not endangered. In so doing they give
recognition to the need to signal solidarity with and/or power over their hear¬
ers, both of which determine the nature of their relations with them. Brown
and Levinson (1978) have developed a model of politeness, in which they dis¬
tinguish a number of options or ‘strategies’ available to the speaker (see Fig¬
ure 5.1). First, the speaker can choose to perform the act or not to perform it.
If the act is performed it can be ‘off-record’ (i.e. performed in such a way that
it can be ignored by the addressee) or ‘on-record’. On-record acts can be
 Pragmatic aspects of learner language 161

‘baldly on record’ (i.e. performed by means of a direct speech act) or can in¬
volve ‘face-saving activity’. The latter can take the form of a ‘positive strat¬
egy’ or a ‘negative strategy’. The former involves some kind of attempt to
establish solidarity with the addressee by emphasizing commonality. It is
likely to occur when there is minimal social distance and little power differ¬
ence between the participants. A ‘negative strategy’ involves performing the
act in such a way that deference is shown to the hearer—the aim is to give the
hearer a way out of compliance with the act. It is used when the power differ¬
ence between the participants is considerable. Brown and Levinson, and also
Scollon and Scollon (1983), argue that the strategies shown in Figure 5.1 are
universal, but subsequent research (see Kasper 1990 for a review) suggests
that this may not be the case. As we will see, how L2 learners handle polite¬
ness has attracted the attention of a number of researchers.

— act not chosen

(5)
off record (e.g. ‘I don’t
speech act — have a match.’)
(4) baldly
on record (e.g. ‘Give
act chosen —
me a match.’)
(3) negative
strategy
on record — — (‘Could
you spare
me a
match?’)
face-saving (2)
activity
positive
strategy
(e.g. ‘Let
me have
your
matches,
love.’)
(1)

Figure 5.1: A schematic representation of Brown and Levinson’s (1978) politeness

model

The study of speech acts in interlanguage has concentrated on illocutionary

meanings, or language functions as they are commonly known. The questions
that have been addressed are: (1) To what extent and in what ways do learn¬
ers perform illocutionary acts in the L2 differently from native speakers of the
target language? and (2) How do learners learn to perform different illocu¬
tionary acts? The bulk of the research has been cross-sectional, so little is cur¬
rently known about the second question. However, a few longitudinal and
162 The description of learner language

pseudo-longitudinal studies cast some light on the process by which learners

gradually master the performance of specific illocutionary acts.
We will begin by considering the research methods that have been used to
investigate illocutionary acts in learner language. A survey of the research
that has studied a number of L2 illocutionary acts follows, with a focus on
three particular acts: requests, apologies, and refusals. A number of general
features of learners’ attempts to perform and to acquire illocutionary acts is
then discussed. Finally, although the focus of this chapter is on description, a
number of explanatory factors will be considered.

Research methods for studying illocutionary acts in learner language

Ideally, the study of illocutionary acts in learner language should involve the
collection of three sets of data: (1) samples of the illocutionary act performed
in the target language by L2 learners, (2) samples performed by native speak¬
ers of the target language3, and (3) samples of the same illocutionary act per¬
formed by the learners in their LI. Only in this way is it possible to determine
to what extent learner performance differs from native-speaker performance
and whether the differences are traceable to transfer from the LI. Relatively
few L2 studies, however, have provided such a base of data.
Kasper and Dahl (1991) distinguish data collection methods according to
the modality of the data elicited (perception/comprehension/intuition versus
production) and the degree of control over learners’ speech (elicited versus
observational). They also point out that some of the most successful studies
have employed combined methods of data collection.
A number of researchers have used questionnaires to elicit learners’ intu¬
itions about how to perform specific acts appropriately in different situations.
Olshtain and Blum-Kulka (1985), for example, investigated learners’ intu¬
itions of politeness in Hebrew by means of a questionnaire consisting of de¬
scriptions of situations calling for apologies and requests followed by lists of
possible strategies for performing these acts in each situation. The learners
were asked to rate each strategy for politeness on a three-point scale. Here is
an example:

A friend of yours at the university comes up to you after class and tells you
that she has finally found an apartment to rent. The only problem is that
she has to pay $200 immediately and at present she only has $100. She
turns to you and says:
a Say, could you lend me $100 until next week?
b You can lend me $100 until next week, right?
c Lend me the money, please,
d Could you lend me $100 until next week?
e Do you think you could lend me $100 until next week?
f Maybe you have a little money to give me so that I could take the
apartment?
(Olshtain and Blum-Kulka 1985: 324)
 Pragmatic aspects of learner language 163

Observational performance data have also been used to investigate the com¬
prehension of illocutionary acts. Carrell (1981), for example, had learners
listen and react to requests, while Kasper (1984a) gauged learners’
comprehension on the basis of the kinds of responses that they provided to
their interlocutor’s previous turn.
The study of learners’ production of illocutionary acts has made use of (1)
discourse completion tasks, (2) role play, and (3) naturally occurring speech.
Discourse completion tasks have been extensively used. In the Cross-Cultural
Speech Act Realization Project (Blum-Kulka, House, and Kasper 1989a) a
series of studies involving subjects from a variety of language backgrounds
(for example, American, British, and Australian English, Canadian French,
Hebrew, German, and Danish) made use of a questionnaire consisting of
eight request and eight apology contexts. Each context was briefly described
and was then followed by a short dialogue with an empty slot which the learn¬
ers were asked to fill in by writing down the request or apology they would
make. Here is an example, designed to elicit an apology:

A student has borrowed a book from her teacher, which she promises to re¬
turn today. When meeting her teacher, however, she realizes that she forgot
to bring it along.

Teacher: Miriam, I hope you brought the book I lent you.

Miriam: _
Teacher: OK, but please remember it next week.
(Blum-Kulka, House, and Kasper 1989b: 14).

Similar questionnaires have been used by Beebe and her colleagues (for ex¬
ample, Beebe, Takahashi, and Uliss-Weltz 1990), and by a number of other
researchers.
Role plays also provide the learners with a description of a context calling
for the performance of a particular illocutionary act. But in this case the learn¬
ers are asked to respond orally. The role plays may be performed with the
help of puppets (for example, Walters 1980), or by the learners interacting
with other learners (for example, Kasper 1981), or with the researcher. The
data collected from role plays provide information about learners’ ability to
construct a discourse context for the specific act under investigation.
The use of naturally occurring speech as a basis for studying interlanguage
pragmatics has been less common, partly because of the difficulty of assemb¬
ling a sufficient corpus of data. Wolfson (1989b), however, used this ap¬
proach to investigate learners’ complimenting behaviour (compliments
together with compliment responses). Bardovi-Harlig and Hartford (1990)
used data collected from academic advising sessions to investigate differences
between native speakers’ and L2 learners’ suggestions.
164 The description of learner language

Each of these methods has its advantages and disadvantages. Controlled

methods such as the discourse completion questionnaire allow for large
amounts of data to be collected quickly, provide information about the kinds
of semantic formulas that learners use to realize different illocutionary acts,
and reveal the social factors that learners think are important for speech act
performance. However, a number of studies which have compared data ob¬
tained from discourse completion questionnaires with that from observa¬
tional studies (for example, Beebe and Cummins 1985; Rintell and Mitchell
1989; Wolfson, Marmor, and Jones 1989) have found differences with re¬
gard to the actual wording used, the semantic formulas employed, the length
of learners’ responses, and the siz.e of the discourse context created. These dif¬
ferences raise questions about the extent to which the elicited data can serve
as evidence of learners’ pragmatic competence, as they may not accurately re¬
flect actual language use. Also, as Bonikowska (1988) has pointed out, in nat¬
urally occurring contexts speakers always have the option of ‘opting out’,
whereas discourse completion questionnaires oblige learners to perform lin¬
guistically even when they would normally keep quiet. Wolfson (1989a) has
also argued strongly that learners’ intuitions about what they would say in a
particular situation are not reliable, as the sociolinguistic knowledge they
draw on in performing illocutionary acts lies beneath the threshold of con¬
sciousness. On the other hand, it is difficult to obtain a sufficient corpus of
data from ethnographic observation. There is also a danger of the data being
unrepresentative of the population under investigation. For example, work
by Wolfson (1983) and Holmes (1986) on the compliments produced by nat¬
ive speakers of English in America and New Zealand was based on data col¬
lected from predominantly female, graduate students. Also, if a
pen-and-paper approach is used, as in some of Wolfson’s early work, it is dif¬
ficult to obtain reliable information about the full discourse context of spe¬
cific illocutionary acts. Kasper and Dahl (1991) conclude that researchers are
‘caughtja&tween a rock and a hard place’.
The study of pragmatics in learner language, then, faces an additional
problem—the matter of target-language norms. In order to decide in what
way the learner’s performance differs from the native speaker’s, it is necessary
to determine what is normative in the latter. We need to know, for example,
how native speakers handle compliments. The difficulty is that native speak¬
ers are likely to vary considerably in this respect. For example, middle-class
Americans have different norms for complimenting from those of middle-
class white South Africans (Herbert 1989). Similarly, females do not compli¬
ment in exactly the same way as males (see Holmes 1988). As Kasper (1992)
points out, the obvious solution to this problem is to choose those norms
found in whatever variety the learner is typically exposed to as the ‘target’.
However, as Kasper also points out, many learners are exposed to more than
one variety. Kasper’s conclusion is that ‘investigators might want to give
more attention to “choosing the right stuff” as an L2 norm in mterlanguage
 Pragmatic aspects of learner language 165

pragmatics’ and that this is ‘largely a matter of logistics’ (1992: 225). In many
learning situations, however, the ‘logistics’ may be very complicated.

Illocutionary acts in learner language

The knowledge required to perform illocutionary acts constitutes part of
communicative competence. Canale includes it in sociolinguistic compet¬
ence, which he defines as ‘the extent to which utterances are produced and un¬
derstood appropriately in different sociolinguistic contexts’ (1983: 7). He
goes on to point out that appropriateness involves both appropriateness of
meaning (i.e. when it is proper to perform a particular illocutionary act) and
appropriateness of form (i.e. the extent to which a given act is realized in a
verbal or non-verbal form proper for a given situation). A similar distinction
is to be found in discussions of pragmatic errors in learner language. Thomas
(1983) distinguishes sociopragmatic failure, which takes place when a learner
fails to perform the illocutionary act required by the situation (i.e. deviates
with regard to appropriateness of meaning), and pragmalinguistic failure,
which occurs when a learner tries to perform the right speech act but uses the
wrong linguistic means (i.e. deviates with regard to appropriateness of form).
A good example of sociopragmatic error can be found in Wolfson’s
(1989a) account of how L2 learners typically respond to compliments. Wolf-
son argues that compliments are used by native speakers of American English
as a means of establishing and maintaining solidarity. It is for this reason that
they are most common among status-equal acquaintances and co-workers
rather than among intimates; the former involve more uncertain relationships
which have to be negotiated. Compliments serve as one of the ways in which
Americans (particularly women) undertake this negotiation. Wolfson points
out that many negotiating sequences involving native speakers are long and
elaborate. In comparison, those involving non-native speakers are typically
short, because learners often fail to take up a compliment, preferring instead
to give no response at all:
NS: You have such a lovely accent.
NNS: (No response).
Wolfson argues that by failing to conform to native-speaker complimenting
norms, learners deprive themselves of the opportunities to establish relation¬
ships with native speakers and, thereby, of the input that they need to develop
both their linguistic and sociolinguistic competence.
Not all of Wolfson’s learners manifested sociopragmatic failure through
failing to respond to a compliment. Many displayed pragmalinguistic fail¬
ure—that is, they responded to a native-speaker compliment but in lin¬
guistically inappropriate ways. Middle-class, white Americans are likely to
respond by giving unfavourable comments about the object that is the target
of a compliment:
166 The description of learner language

NS: I like your sweater.

NS: It’s so old. My sister bought it for me from Italy a long time ago.

In contrast, learners often tried to refuse the compliment or to downgrade

themselves. They also tended to respond with a simple ‘thank you’. Such re¬
sponses also served to act as dampeners on the conversation.
Pragmalinguistic failure by learners is widely reported in the literature. An¬
other good example comes from Eisenstein and Bodman’s (1986) study of ex¬
pressions of gratitude. This made use of a discourse completion questionnaire
administered to 67 learners with different LI backgrounds. Baseline data
were collected from native speakers of English. Eisenstein and Bodman report
that the learners performed very differently from the native speakers—in
fact, their responses were similar only 30 per cent of the time. They experi¬
enced difficulty with both syntax and vocabulary, and also in identifying the
formulas and conventionalized routines that characterized native-speaker
thanking. In a subsequent study (Bodman and Eisenstein 1988), learners were
asked to role play situations calling for expressions of gratitude. Lower-
proficiency learners often translated expressions from their LI. For example,
in thanking someone for a loan they might say ‘May God increase your
bounty’. More advanced learners avoided this kind of pragmalinguistic error,
but instead displayed considerable hesitation and awkwardness. Prag¬
malinguistic difficulty, therefore, can also be manifested in the failure to con¬
form to the temporal norms of native-speaker speech.4
The distinction between sociopragmatic and pragmalinguistic errors, how¬
ever, is not as clear-cut as these examples may have suggested. Kasper points
out that ‘the distinction becomes fuzzy in the case of indirectness’ (1992:
210). For example, the decision regarding whether to provide an explanation
for having committed some offence can be seen as a sociopragmatic one, but
if providing an explanation is seen as one of several possible strategies for per¬
forming the act of apologizing (as in Olshtain and Cohen’s 1983 framework
discussed later), its inclusion or omission constitutes more of a prag¬
malinguistic decision. Nevertheless, Kasper acknowledges that the distinc¬
tion is ‘analytically useful’.
Research into the use and acquisition of illocutionary acts has been some¬
what limited. It has tended to make use of rough and ready categories of
sociocultural reality—the problem of norms referred to earlier. It has also
tended to concentrate on a fairly small set of speech acts. Complaints have
been investigated by Olshtain and Weinbach (1987) and Bonikowska (1988),
thanking by Eisenstein and Bodman (1986), invitations by Scarcella (1979),
suggestions by Bardovi-Harlig and Elartford (1990) and Rintell (1981), com¬
pliments by Wolfson (1989b) and Billmyer (1990), and arguing by Adger
(1987). Many of these acts have two points in common. First, they constitute
‘relatively well-defined’ acts (De Beaugrande and Dressier 1981: 117) in the
sense that they are realized by means of a small set of easily recognizable
 Pragmatic aspects of learner language 167

linguistic elements (many formulaic). For example, Wolfson (1983) has

shown that nine syntactic patterns account for 95 per cent of the compliments
in her native-speaker corpus. Second, these acts are face-threatening in nature
and, therefore, provide a means of studying to what extent L2 learners with
different LI backgrounds are able to use native-like politeness strategies. We
will focus on three acts (requests, apologies, and refusals) which are among
those which have received the most attention from researchers.

Requests

Requests are attempts on the part of a speaker to get the hearer to perform or
to stop performing some kind of action. A number of general interactional,
illocutionary, and sociolinguistic features of requests can be identified:

Interactional features

1 They tend to serve an initiating function in discourse.

2 They can be performed in a single turn or, if they involve some kind of
preparatory act or pre-request, over several turns, for example:
T: Have you finished your work?
S: Yes.
T: Read your book then.

Illocutionary aspects

3 The speaker wishes the hearer to perform the request, believes the hearer is
able to perform the act, and does not believe the act will be performed in
the absence of the request (Fraser 1983).
4 A request can be more or less direct (Searle 1976). Blum-Kulka, House,_gnd
Kasper (1989b) identify eight ‘strategy types’, which they order according
to directness (see Table 5.1).
5 Requests are also subject to internal and external modification. Internal
modification takes the form of downgraders, which are intended to mitig¬
ate the force of the act, and upgraders, which are intended to increase the
degree of coerciveness of the act. External modification consists of moves
that occur either before or after the head act (i.e. the act that actually per¬
forms the request); these moves can also be classified according to whether
the purpose is to downgrade or upgrade the force of the act.
6 Requests can be encoded from the speaker’s perspective (for example,
‘Give me the book’), from the hearer’s perspective (for example, ‘Could
you give me the book?’), from a joint perspective (‘Let’s read a book’) or
from an impersonal perspective (‘It would be nice to read a book’).
168 The description of learner language

Sociolinguistic aspects

7 Requests are ‘inherently imposing’ (Blum-Kulka, House, and Kasper

1989b). For this reason they call for considerable ‘face-work’. The choice
of linguistic realization depends on a variety of social factors to do with the
relationship between the speaker and the addressee, and the perceived de¬
gree of imposition which a particular request makes on the hearer (i.e. it
involves a choice of politeness strategy).
8 Although the main sociopragmatic categories of requests can be found in
different languages, there are pragmalinguistic differences relating to the
preferred form of a request that is used in a particular situation. Also,
cross-linguistic differences exist in the choice of other linguistic features
such as internal and external modification devices.

Level of directness Strategy Example

Direct 1 Mood-derivable You shut up.

Conventionally indirect
Non-conventionally indirect
2 Performative 1 am telling you to shut up.
3 Hedged performative 1 would like to ask you to
shut up.
4 Locution-derivable 1 want you to shut up.
5 Suggestory formula Let’s play a game.
6 Query-preparatory Can you draw a horse for
me?
7 Strong hint This game is boring.
8 Mild hint We’ve been playing this
game for over an hour now.

Table 5.1: Request strategies (summarized from Blum-Kulka, House, and Kasper
1989b)

Requests have received considerable attention in SLA research for a num¬

ber of reasons. They are face-threatening and, therefore, call for considerable
linguistic expertise on the part of the learner, they differ cross-linguistically in
interesting ways and they are often realized by means of clearly identifiable
formulas. Table 5.2 summarizes the main studies.
A number of studies have investigated learners’ intuitions about what con¬
stitutes an appropriate request. A focus of enquiry is whether L2 learners are
able to recognize the distinctions between polite and less polite forms. Ad¬
vanced learners appear to have few problems. Olshtain and Blum-Kulka
(1985) found that learners of L2 Hebrew, who had been resident in Israel for
over 10 years, showed the same high level of tolerance for direct and positive
politeness strategies as native speakers of Hebrew. Walters (1979) and
Carrell and Konneker (1981) also report that the advanced learners they
 Pragmatic aspects of learner language 169

Study Type Modality Subjects Data collection Main results

studied instruments

Scarcella 1979 Cross-sectional Production 10 beginners & 10 Role play involving Some politeness
advanced learners different addressees forms (e.g. ’please’
(LI Arabic); 6 NSs acquired early;
others (e.g. inclusive
’we') acquired late.
Learners acquire
forms before their
social meanings. L2
learners use a
limited range of
politeness features.
Walters 1979 Cross-sectional Perception 75 advanced ESL Judgement tasks Learners performed
learners (mixed Lis); involving paired like NS females in
60 NSs of English comparisons. making clearer
distinctions
between polite and
impolite forms than
NS males.
Fraser, Rintell, and Cross-sectional Perception Adult ESL learners Ratings of speech Learners perceived
Walters 1980 (LI Spanish); NS produced in role- deference to older
controls playing situations on and female
a deference scale of addressees in a
1-5. similar way to NSs.
(i.e. they did not
transfer LI
pragmatic rules).
Carrell and Cross-sectional Perception 73 advanced and Sorting task, Learners performed
Konneker 1981 intermediate ESL requiring subjects to like NSs but tended
learners (mixed Lis); rank strategies to perceive more
42 NSs according to distinctive levels of
politeness. politeness i.e. they
were oversensitive
to syntactic/form
distinctions.
Tanaka and Kawade Cross-sectional Perception ESL learners (LI M/C questionnaire; Both learners and
1982 Japanese) NS subjects select NSs choose more
controls strategy most polite strategies with
appropriate to increased social
situation. distance but
learners chose less
polite strategies
overall.
Schmidt 1983 Longitudinal (3 Production One adult ESL Audio recordings in Learner showed
yrs) learner (LI different natural initial reliance on
Japanese) settings small set of formulas
and lexical clues
(’please'); by end he
had control of wide
range of request
formulas but could
not always use them
appropriately.

(continued overleaf)
170 The description of learner language

Study Type Modality Subjects Data collection Main results

studied instruments

Olshtain and Blum- Cross-sectional Perception 172 NSs of M/C questionnaire Learners showed
Kulka 1985 English; 160 NSs in English and increased
of Hebrew; 124 Hebrew; subjects preference for
learners of Hebrew asked to rate each direct request
of 6 strategies for strategies and
politeness on a positive politeness
3-point scale (i.e. NS norms)
according to length
of stay in target
community;
learners capable of
achieving NS
acceptability
patterns.
Cathcart 1986 Cross-sectional Production 8 ESL learners Audio-recordings Children used
(children; LI of classroom longer and more
Spanish) in speech complex requests
bilingual classroom addressed to
adults than to other
children. Requests
also longer and
more complex with
other children in
tasks with joint
goals.
Blum-Kulkaand Cross-sectional Production Learners of L2 Discourse High-intermediate
Olshtain 1986 Hebrew; (mixed completion learners used
proficiency); NS questionnaire longer requests
controls than low-
intermediate or
advanced learners.
Ervin-Trippetal. Cross-sectional Comprehension Learners of French Subjects asked to The child learners
1987 (3 age groups 3-9); explain implicit relied on situational
NSs of English and requests based on rather than
French narratives; also to linguistic clues to
carry out implicit interpret requests;
requests. accuracy of
interpretation
increased with age.
House and Kasper Cross-sectional Production Advanced learners Discourse Learners showed
1987 of L2 English (LI = completion similar choice of
Danish and questionnaire directness levels
German) but used less
varied syntactic
and lexical down-
graders less
frequently. Also,
r
learners produced
longer requests.
Tanaka 1988 Cross-sectional Production 8 Japanese ESL Role play involving The learners were
learners in two hearers (friend more direct and
Australia; NS and lecturer) used politeness
controls strategies
inappropriately.
 Pragmatic aspects of learner language 171

Study Type Modality Subjects Data collection Main results

studied instruments

Faerch and Kasper Cross-sectional Production 200 learners of Discourse Learners tend to be
1989 English & 200 completion more verbose e.g.
learners of German; questionnaire tend to use double
(LI = Danish) NS of markings (‘Could
Danish, British you possibly...?’)
English and German ‘Please’ is over¬
used, but ‘possibly’
is underused.
Rintell and Mitchell Cross-sectional Production 34 ESL learners Discourse Learners’ requests
1989 (low-advanced level, completion longer than NSs’; no
mixed Lis); 37 NSs questionnaire; role major differences in
play choice of forms/
strategies.
Ellis 1992a Longitudinal (2 Production 2 ESL learners Pencil and paper Clear evidence of
yrs) (10/11 yrs old; Lis records and audio developmental
Portuguese and recordings progression but
Punjabi) classroom learners did not
setting make elaborated
requests and were
limited in their ability
to use requests in a
number of ways
(e.g. used few hints
and little
modification).

Table 5.2: Survey of studies of interlanguage requests

studied perceived the politeness level of different requests in accordance with

native-speaker norms. However, there is also evidence of some differences. In
particular, advanced learners appear to develop a greater sensitivity to the
use of politeness strategies in requesting than is evident in native speakers.
Walters notes that there was greater unanimity between his learners and
native-speaker females, who tended to distinguish polite and impolite forms
more sharply than native-speaker males. Carrell and Konneker found that
their advanced learners tended to perceive more distinct levels of politeness
than native speakers. A tentative conclusion, therefore, is that with sufficient
exposure to the L2, learners are able to perceive the sociolinguistic distinc¬
tions encoded by native speakers in requests, but that they may become over¬
sensitive to them.
There is less information available about how low-level (as distinct from
advanced or intermediate) learners understand different kinds of requests.
One possibility is that such learners rely more on situational than linguistic
clues as a means of comprehending requests, as Ervin-Tripp et al.’s (1987)
study of child learners of L2 French suggests. Low-level learners are likely to
have problems selecting request strategies that are appropriate to different
situations. Thus, Tanaka and Kawade (1982) found that Japanese ESL learn¬
ers (who had probably had little exposure to the sociolinguistic norms of nat¬
ive speakers) tended to opt for less polite strategies overall. Further
172 The description of learner language

evidence of the problems that learners face in their choice of request strategy
is to be found in Olshtain and Blum-Kulka (1985). They noted that learners
of L2 Hebrew who had spent less than two years in Israel were reluctant to ac¬
cept the informal, positive-orientated request strategy or the direct request
strategy found in native-speaker speech. For example, learners were inclined
to reject the Hebrew equivalent of ‘I hope you can take me back to town’
when asking for a ride, whereas native speakers were more likely to accept it.
The learners may have responded in accordance with the politeness norms for
requesting of their LI.
Most of the research has focused on the production of requests elicited in
the form of written responses to a discourse completion questionnaire or oral
responses to role play tasks. Many of the studies have investigated high-
intermediate or advanced learners. One of the strongest findings from these
studies is that even these learners do not acquire fully native-like ways of re¬
questing. In particular, they tend to produce longer requests than native
speakers—the ‘waffle phenomenon’ (see Blum-Kulka and Olshtain 1986;
House and Kasper 1987; Fasrch and Kasper 1989; Rintell and Mitchell 1989;
Edmondson and House 1991). This is the result of the over-suppliance of po¬
liteness markers, of syntactic downgraders, and, in particular, of supportive
moves. This verbosity is more evident in high-intermediate than in advanced
learners (Blum-Kulka and Olshtain 1986). A number of explanations have
been suggested for this phenomenon. It may reflect a desire to ‘play it safe’ by
making propositional and pragmatic meanings as transparent as possible.
The need to ‘play it safe’ may in turn derive from the fact that although ad¬
vanced learners have access to the standardized routines needed to perform
requests, they make less use of them than native speakers because they are not
sure about the range and appropriateness of their application in particular
situations (Edmondson and House 1991: 285). Verbosity may also reflect a
desire on the part of such learners to display their linguistic competence or,
conversely, it may be indicative of the learners’ awareness of their own status
as second language users—an acknowledgement of the diminished identity
they feel in certain situations. Again, though, conclusions about learners’
pragmatic competence on the basis of their elicited performance can only be
tentative.
Other differences between high-level learners and native speakers have
been noted. For example, they tend to overuse the politeness marker ‘please’,
to employ more double-markings (for example, ‘Could you possibly present
your paper this week?’) and to make more extensive use of external (as op¬
posed to internal) modification (Faerch and Kasper 1989). These high-level
learners, however, show control over a wide range of forms and strategies for
performing requests.
In contrast, lower-level learners display only a limited range of politeness
features. Scarcella (1979), for instance, reports that ‘low level L2 speakers
showed much less variety in the politeness strategies they used and lacked any
 Pragmatic aspects of learner language 173

apparent distribution of politeness features’ (page 287). For example, they

used imperative requests to all addressees irrespective of social distance and
power differences, and deployed hedges in totally inappropriate ways.
Tanaka (1988) found that Japanese students at an Australian university dif¬
fered markedly from native English speakers in their requesting behaviour.
They failed to give concrete reasons for their requests and also showed less
uncertainty regarding whether the conditions for a request were met. In gen¬
eral, they were less indirect and tentative. Thus, whereas Australians were
likely to say ‘Do you think I could have ...?’ when requesting a book from a
lecturer, the Japanese used ‘Can I ...?’ They also differed in their choice of
strategies, using negative politeness strategies (for example, ‘If you don’t
mind ...’) where the native speakers did not, and failing to use them in other
situations where the native speakers employed them. They avoided positive
politeness strategies (for example, mention of first name). These lower-level
learners had no difficulty in performing the illocutionary meaning of re¬
questing, but they were unable to do so in socially appropriate ways. They
made an attempt to vary their request in accordance with social factors such
as addressee—as Cathcart (1986) also found in a study of the communicative
behaviour of children in a bilingual kindergarten—but could not do so very
effectively.
Ideally, detailed information is needed about how individual learners with
different LI backgrounds gradually develop the ability to perform requests
over time. Unfortunately, there have been few longitudinal studies so far.
Schmidt’s (1983) study of Wes examined his acquisition of directives (a gen¬
eral class of speech acts that includes requests) over a three-year period. Ini¬
tially, Wes relied on a small set of formulas (for example, ‘Can I have a ...?’).
On occasions he used V + -ing in place of the imperative form, and he tended
to rely heavily on lexical clues such as ‘please’ (a phenomenon that Scarcella
also noted in beginner learners). Wes also made substantial use of hints,
which Schmidt says native speakers often had difficulty in interpreting. Over
three years considerable development took place so that ‘by the end of the
period gross errors in the performance of directives had largely been elimin¬
ated’ (1983:154). Wes could use request formulas productively, he had aban¬
doned V + -ing in commands, he had increased the range of request patterns at
his disposal (for example, ‘Let’s ...’ and ‘Shall we ...?’ had appeared), and in
general gave evidence of much greater elaboration in his requesting behavi¬
our. However, Schmidt notes that Wes was still limited in his ability to vary
the use of directive type in accordance with situational factors, and also some¬
times used a particular pattern in inappropriate ways.
The second longitudinal study indicates a similar pattern of development
to that of Wes, although in this case the two learners were children acquiring
L2 English in a classroom context in London. In Ellis 1992a, I claimed that
there was clear evidence of a developmental progression. For example, re¬
quests for goods were first performed by means of verbless utterances (such as
174 The description of learner language

‘Pencil please?’). Next, mood derivable requests with an imperative verb

appeared (such as ‘Give me a paper’) and only a short time afterwards query
preparatory requests using ‘Can I have ... ?’ This formula was frequently used
and alternative ways of requesting goods did not appear until much later. It
also took time for ‘Can I...?’ to appear in a range of syntactic frames (for ex¬
ample, ‘Can I take a book with me?’ and ‘Can you pass me the pencil?’). I also
identified a number of general areas of development. For example, the overall
range of request types expanded, the range of exponents of a specific request
type increased, and requests that encoded the hearer’s perspective (as op¬
posed to the speaker’s) emerged. However, the learners’ requests were still
limited in a number of ways by the end of the two-year period. The majority
of their requests were still of the direct kind. There were few examples of non-
conventional requests. Certain types of requests (performatives and hedged
performatives) did not occur at all. The range of formal devices was still very
limited. There was little attempt to employ either internal or external modi¬
fication. The learners failed to modify their choice of request strategy system¬
atically according to addressee (teacher or another pupil). I suggested two
possible explanations for these developmental limitations. One was that the
learners were still in the process of acquiring the linguistic and pragmatic
knowledge needed to perform requests. According to this explanation, many
features of requests are acquired late. The other explanation was that the
classroom setting did not afford the appropriate communicative conditions
for full acquisition of request forms and strategies, a point taken up later in
Chapter 13.
What general conclusions can we come to regarding L2 learners’ use and
acquisition of requests? Learners—even relative beginners—appear to have
few problems in understanding the illocutionary force of a request, probably
because they are able to make use of situational cues. Later on they are able to
perceive the sociolinguistic meanings encoded by different request types, al¬
though they tend to be oversensitive to these. With regard to production,
learners begin with very simple requests and then slowly build up their reper¬
toire, learning not only an increasing number of formal devices for per¬
forming them, but also the social meanings attached to these devices. There is
considerable evidence, however, to suggest that even very advanced learners
fall short of native-speaker competence. Although they eventually master
native-speaker politeness strategies, they show a tendency to verbosity, per¬
haps because they are aware of the dangers that are inherent in making
requests.

Apologies

Like requests, apologies are face-threatening acts. However, they differ from
requests in. that they impose on the speaker rather than the hearer. An
apology requires the speaker to admit responsibility for some behaviour (or
 Pragmatic aspects of learner language 175

failure to carry out some behaviour) that has proved costly to the hearer.
Apologies also differ from requests in that they refer to past rather than future
events.
Apologies may also differ from requests in another important way.
Whereas there are substantial cross-cultural differences in the way requests
are realized in different situations, this does not appear to be the case with
apologies. Olshtain (1989), in a study that was part of the Cross-cultural
Speech Act Realization Project, investigated the strategies used to realize
apologies by speakers of four different languages: Hebrew, Australian
English, Canadian French, and German. As in most of the other studies in this
project, data were collected by means of a discourse completion task.
Olshtain reached the following conclusion:

... we have good reason to expect that, given the same social factors, the
same contextual factors, and the same level of offence, different languages
will realize apologies in very similar ways (1989: 171).

In other words, the strategies used to perform apologies are largely universal.5
As such, it might be expected that this speech act will pose L2 learners few
problems, as they will be able to make use of LI strategies. However, as we
will see, learners do experience difficulty in apologizing in an L2.
The principal strategies for apologizing have been described by Olshtain
and Cohen (1983). They point out that when a speaker is confronted with a
situation in which the interests or rights of a hearer have been violated, one of
two things can happen: the speaker can reject the need for an apology, or can
accept responsibility for the violation and apologize. In the case of the former,
the speaker may deny that there is a need to apologize or deny responsibility
for the violation. In the case of the latter, there are a number of strategies
which the speaker can choose from and perform with different levels of
intensity.
The basic strategies, together with examples of the semantic formulas that
realize them, are shown in Table 5.3. Following Fraser (1981), a ‘semantic
formula’ is defined as ‘a word, phrase or sentence which meets a particular se¬
mantic criterion or strategy’ (cited in Olshtain and Cohen 1983: 20). The
strategies and formulas shown in Table 5.3, which comprise the speech act set
for apologies, have been confirmed in subsequent studies. Blum-Kulka,
House, and Kasper (1989b), for instance, provide a similar list. The strategies
vary in their directness. The first (‘an expression of an apology’), which is
realized by an explicit illocutionary force indicating device of a formulaic na¬
ture, constitutes a direct strategy, whereas the other strategies are more indir¬
ect. An apology can be performed with different levels of intensity. For
example, the speaker can say ‘I’m sorry’ or ‘I’m very sorry’, or can use mul¬
tiple strategies to increase the intensity. There are also strategies for down¬
grading apologies, such as when the speaker says ‘Sorry, but you shouldn’t be
so sensitive’ (Blum-Kulka, House, and Kasper 1989b: 21). It should also be
176 The description of learner language

noted that paralinguistic features such as tone of voice are likely to be

important in apologizing. However, these have not been systematically
investigated.

Strategy Example

1 An expression of an apology
a expression of regret I’m sorry.
b an offer of apology 1 apologize.
c a request for forgiveness Excuse me.
2 An explanation or account of the situation The bus was late.
3 An acknowledgement of responsibility
a accepting the blame It’s my fault.
b expressing self-deficiency 1 wasn’t thinking.
c recognizing the other person as
deserving apology You are right.
d Expressing lack of intent 1 didn’t mean to.
4 An offer of repair I’ll pay for the broken vase.
5 A promise of forbearance It won’t happen again.

Table 5.3: The speech act set for apologies (information and examples taken from
Olshtain and Cohen 1983)

Apologies in an L2 have been studied by Cohen and Olshtain (1981),

Olshtain and Cohen (1983), Fraser (1981), Olshtain (1983), Blum-Kulka and
Olshtain (1984), Olshtain and Cohen (1989), and Rintell and Mitchell
(1989). All these studies made use of data elicited by means of either a written
discourse completion or oral role play task. The subjects of these studies came
from a variety of language backgrounds. In most of them baseline data from
the learners’ LI and from native speakers of the target languages were also
collected. The studies suggest that L2 learners’ performance of apologies is
influenced by a number of factors: (1) the learners’ level of linguistic
proficiency, (2) their LI, (3) their perception of the universality or language
specificity of how to apologize, and (4) the nature of the specific apology
situation.
Evidence that the learners’ level of linguistic proficiency influences how
they apologize is forthcoming when it can be shown that subjects’ LI apol¬
ogies and those of native speakers of the target language are very similar, but
differ from the subjects’ L2 apologies. In such cases LI transfer is possible but
does not actually occur, presumably because the learners lack the necessary
L2 proficiency to do so. Cohen and Olshtain (1981) provide several examples
in a study of Hebrew learners of L2 English. They found situations where the
learners did not seem to be familiar with the semantic formulas needed for the
apology. For example, they were less likely to offer repair when they had
backed into someone’s car, and less likely to acknowledge responsibility
when they had bumped into and shaken up an old lady than when they
 Pragmatic aspects of learner language 177

performed the same apologies in their LI. Lack of linguistic proficiency was
in this case reflected in the use of general formulas and saying too little.
In a later paper, Olshtain and Cohen (1989) distinguish three types of devi¬
ation resulting from gaps in linguistic competence. Overt errors occur when
the learner is evidently trying to apologize but produces a linguistic error as in
this example:

Situation: bumping into a woman in the way. ‘I’m very sorry but what can I
do? It can’t be stopped’.

where the speaker uses ‘stopped’ instead of ‘avoided’. Covert errors occur
when the learner’s apology is linguistically correct but is inappropriate. For
example, a Hebrew learner of L2 English apologized for failing to keep a
meeting with a friend with ‘I really very sorry. I just forgot. I fell asleep. Un¬
derstand’. Faulty realization of a semantic formula occurs when the learner
has chosen an appropriate formula but phrases it incorrectly, as when one
learner wanted to offer repair for forgetting a meeting with someone by using
this formula: ‘I think I can make another meeting with you’.
Despite the fact that all languages share the same set of basic strategies for
apologizing, negative transfer is evident in a number of respects. For example,
the learner’s LI may have an effect on the intensity with which apologies are
performed. Cohen and Olshtain (1981), for example, note that Hebrew
speakers of English were less likely to accept responsibility for an offence or
to make offers of repair than native English speakers. Also, they did not in¬
tensify their expressions of regret as much. Olshtain and Cohen (1983) refer
to an unpublished study by Wu (1981) which found that Chinese learners of
English intensified regret much more than native English speakers. These dif¬
ferences corresponded to differences in the learners’ Lis. Olshtain and Cohen
comment ‘while Hebrew L2 speakers may appear somewhat rude to native
English speakers when expressing regret, Chinese L2 speakers may appear
overly polite, even obsequious’ (1983: 30). The Chinese learners also tended
to offer explanations, where native English speakers did not. This resulted in
the same phenomenon observed in studies of L2 requests—a tendency to say
too much.
However, it may not be the actual differences between languages, but
rather the learners’ attitudes about how apologies should be performed cross-
linguistically, that matter. Olshtain (1983) found that the overall frequency
of semantic formulas was higher in native English speakers than in native
Hebrew speakers, with native Russian speakers intermediate:

English > Russian > Hebrew

She hypothesized that ‘transfer from Russian or English into Hebrew as the
target L2 might therefore be expected as an increase in the overall frequency
of use of semantic formulas’ (1983: 245). However, this was not quite what
she found. Whereas the English learners decreased the frequency of their use
178 The description of learner language

of semantic formulas to a level approximating to that of native Hebrew

speakers, the Russian learners maintained the same level as in their native
Russian. Olshtain suggests that this reflects the different perceptions of the
English and Russian learners. Whereas the former perceived Hebrew speak¬
ers as apologizing less than native English speakers, the latter viewed apolo¬
gizing as a universal phenomenon, ‘claiming that people needed to apologize
according to their feelings of responsibility, regardless of the language which
they happen to be speaking...’ (op. cit.: 246). Thus, transfer was governed by
whether the learners saw apologies as language-specific or universal in na¬
ture. However, as Kasper (1992) notes, the Russian learners in fact supplied
apology strategies in three out of-five situations more frequently in L2 Heb¬
rew than in their native Russian, despite their perceptions of universality.
This suggests that learners’ attitudes can be overridden by other factors.
Olshtain (1983) also argues that transfer effects can only be accurately
identified if attention is given to the specific situation that leads to an apology.
She suggests that transfer might be more likely in situations where the learn¬
ers’ culturally-determined perceptions of the importance of status and dis¬
tance and of the severity of the offence are different from those of native
speakers of the target language. For example, in the case of a situation involv¬
ing backing into someone’s car and causing damage, English speakers of L2
Hebrew apologize in much the same way as they do in their LI, whereas in the
situation where they have insulted someone at a meeting, they do not transfer
their LI strategies but rather behave in a similar way to native Hebrew speak¬
ers. The extent to which learners acquire the sociocultural rules of the L2 is,
therefore, situation-dependent.
To sum up, the study of L2 apologies bears out many of the findings of the
research on L2 requests. Both the learners’ general level of linguistic profi¬
ciency and the socio-cultural norms of their LI influence how they apologize
in an L2. Learners may be too concise or they may be verbose. The extent to
which transfer takes place can be influenced by the learners’ perceptions of
the pniversality of how to apologize, transfer being less likely if learners re¬
cognize the language-specificity of apologies. Transfer is also more likely in
situations where learners feel the need to act in accordance with the sociocul¬
tural norms of the native culture. The research to date has not examined how
learners of different levels of proficiency perform apologies, or how apologies
are learnt over time.

Refusals

Refusals constitute another face-threatening act; in this case they threaten the
hearer more than the speaker. They require a high level of pragmatic compet¬
ence. L2 learners’ refusals have been studied in a series of investigations
involving Beebe and her co-researchers (Beebe and Takahashi 1989a and
1989b; Takahashi and Beebe 1987; Beebe, Takahashi, and Uliss-Weltz
 Pragmatic aspects of learner language 179

1990). These studies relied primarily on data collected from a written dis¬
course completion questionnaire, but some specific examples of naturally
occurring refusals are also discussed.
Refusals occur in the form of responses to a variety of illocutionary acts
(for example, invitations, offers, requests, and suggestions). On the basis of
an analysis of native-speaker refusals, Beebe and Takahashi were able to
show that they are performed by means of a fairly limited set of direct and in¬
direct ‘semantic formulas’, which are shown in Table 5.4. Individual refusals
are made up of different selections from these formulas in accordance with
the status and power relationships holding between speaker and hearer. The
analysis of learners’ refusals considered (1) the order of semantic formulas,
(2) the frequency of formulas, and (3) the content of individual semantic for¬
mulas in relation to native-speaker use.

Type Semantic formula Example

Direct 1 Performative 1 refuse.

2 Non-performative statement
Indirect6 3 Statement of regret
4 Wish
5 Excuse, reason, explanation
6 Statement of alternative
7 Set condition for past or future
acceptance
8 Promise of future acceptance
9 Statement of principle
10 Statement of philosophy
11 Attempt to dissuade interlocutor
12 Acceptance that functions as
refusal
13 Avoidance (e.g. silence or
hedging)
1 can’t.
I’m sorry.
1 wish 1 could help you.
1 have a headache.
I’d prefer to...
If you’d asked me earlier I’d have ...
I’ll do it next time.
1 never do business with friends.
One can’t be too careful.
1 won’t be any fun tonight.
Well, maybe.
I’m not sure.

Table 5.4: Semantic formulas used in refusals (based on information provided in

Beebe, Takahashi, and Uliss-Weltz 1990)

The subjects of the various studies were Japanese learners of English. Beebe
and Takahashi are at pains to point out that there are certain stereotypes re¬
garding Japanese people. Japanese people are supposed to apologize a lot, to
be less direct and less explicit than Americans, to avoid making critical re¬
marks to someone’s face, to avoid disagreement, and to avoid telling people
things that they do not want to hear. The results of their research indicate that
these stereotypes are not warranted. Frequently, for instance, the learners
were more direct than the native speakers and in certain situations they
180 The description of learner language

showed no reluctance to impart unpleasant information. There were,

however, a number of differences between the way Japanese learners and
Americans performed refusals.
One type of pragmalinguistic error concerned the order of semantic for¬
mulas. Beebe, Takahashi, and Uliss-Weltz (1990) found that although even
proficient Japanese speakers of English in the United States employed the
same range of semantic formulas as Americans, they differed in the order in
which they were typically used. For example, the Japanese speakers omitted
expressions of apology or regret in refusing invitations made by people lower
in status than themselves. They reacted differently according to whether the
invitation originated from a higher- or lower-status person, whereas the nat¬
ive speakers responded according to how familiar they were with their inter¬
locutors. The same difference was evident in the frequency with which
semantic formulas were used. The Japanese English speakers increased the
number of formulas they used when refusing a higher-status interlocutor,
while the American English speakers did so when addressing familiar equals.
In other words, where Americans adopted strategies consonant with solidar¬
ity, the Japanese preferred power-orientated strategies. A similar difference is
evident in the content of semantic formulas, the Japanese excuses tending to
be less specific than American excuses (except when refusing food) and
sounding more formal in tone. This was particularly evident in the frequent
use of lofty-sounding appeals to principle and philosophy. For example,
refusing the offer of a new diet, one Japanese learner responded ‘I make it a
rule to be temperate in eating’.
The main focus of Beebe and Takahashi’s research is pragmatic trans¬
fer—the extent to which learners transfer the ‘rules of speaking’ of their nat¬
ive language into the L2. A key question is to what extent transfer is
influenced by the learners’ level of L2 proficiency. Takahashi and Beebe
(1987) hypothesized that Japanese learners (EFL and ESL) with higher profi¬
ciency would display more Japanese communicative characteristics in their
English than less proficient learners. They based this hypothesis on the argu¬
ment that lower-level learners lack the linguistic resources to encode socio¬
culturally appropriate Japanese patterns and so resort to a simplification
strategy, whereas the higher-level learners have acquired the linguistic means
to make socio-cultural transfer possible. The results of an analysis of the
refusals produced by Japanese ESL learners gave support to this hypothesis.
The more proficient learners (defined in this study as those who had been res¬
ident longer in the United States) made more frequent use of native-language
patterns—in particular, the high level of formality in the tone and content of
refusals (for example, ‘I am very delighted and honoured to be asked to attend
the party, but...’). However, the results for the EFL learners (i.e. those study¬
ing English at college level in Japan) failed to support the hypothesis, there be¬
ing no difference in the refusals of undergraduates and graduates. Takahashi
and Beebe suggest that this was because ‘pragmatic competence is not
 Pragmatic aspects of learner language 181

affected by just a few years’ difference in school in the EFL context’ (page
149). Again, this study suggests that the development of pragmatic compet¬
ence depends on whether the learners experience any sociolinguistic need to
vary their performance of specific acts.
The study of L2 refusals is more limited than the study of requests or
apologies. There have been fewer studies, none of them longitudinal, and a
narrow range of subjects (only adult Japanese). However, the work done
raises a number of interesting points. First, F2 learners’ pragmatic behaviour
is not always in accordance with stereotypical views. Second, although
advanced F2 learners have no difficulty in performing refusals, they do not
always do so in the same way as native speakers. One possible reason for this
is pragmatic transfer. Thus, highly proficient Japanese learners respond to the
status of their interlocutors, rather than to how familiar with them they are.
Third, learners may need to reach a threshold level of linguistic proficiency
before pragmatic transfer can take place.

Some generalizations regarding the performance of illocutionary acts

in learner language
Although the study of interlanguage pragmatics is still in its infancy (there be¬
ing an unfortunate lack of longitudinal studies), it is possible to identify a
number of general characteristics:

1 Just as learners make linguistic errors, so too they make pragmatic errors.
These are of two kinds: sociopragmatic and pragmalinguistic (Thomas
1983).
2 Following and building on Blum-Kulka (1991), three broad phases of de¬
velopment of pragmatic competence can be identified:

Phase 1: Message-orientated and unsystematic:

During this stage learners rely on situational clues to interpret the
illocutionary force of different acts and on simplification strate¬
gies to produce them (for example, the use of formulas and the
over-use of simple politeness markers such as ‘please’7). The
learner uses ‘any linguistic and non-linguistic means at his or her
disposal to achieve a communicative end’ (1991: 270).
Phase 2: Interlanguage-orientated and potentially systematic:
During this stage learners are able to differentiate the social
meanings performed by alternative realizations of an illocution¬
ary act. They are also able to use a variety of strategies in their
own production. However, their performance will differ from
that of native speakers in two major ways: it is more verbose and
it shows evidence of pragmatic transfer from the native language.
Thus ‘the learners’ speech acts will be pragmatically and socially
acceptable in part and in part unacceptable’ (1991: 270).
182 The description of learner language

Phase 3: Interculturally-orientated and potentially systematic:

During this stage learners approximate closely to native-speaker
performance. Thus, they use the same range of politeness strate¬
gies as native speakers. However, they may continue to display
residual transfer of ‘deep’ cultural elements (for example, the im¬
portance attached to status differences), to use too many words,
to manifest a different distribution of linguistic devices used to
realize speech acts, and, generally, to display oversensitivity and
hesitancy in comprehending and performing face-threatening
acts.

3 It probably takes a large number of years of exposure to the L2, and also a
setting that creates a need for learners to vary their performance of illocu¬
tionary acts in accordance with the addressee and other factors, before
learners reach Phase 3. Many learners will probably never do so.

A number of different factors have been proposed to explain this pattern of

development. Obviously, the learner’s linguistic proficiency is a key one.
Learners with little lexical or grammatical knowledge of the L2 will be re¬
stricted not only in what illocutionary acts they can perform, but also in how
they perform them. In the previous chapter, it was suggested that learners first
acquire forms and later learn what functions they can use them to perform, a
view supported by Scarcella’s (1979) study of requests. For example, she
found instances of forms used at least ten times by individual learners but
without variation according to addressee.
Transfer is the other major factor identified. Learners perform speech acts
such as requests, apologies, and refusals in accordance with the sociolinguis-
tic rules of their native language. However, as discussed above, transfer can
only take place when learners have achieved sufficient L2 resources to make it
possible. Also, the extent to which transfer takes place may be governed by
the learners’ perceptions about transferability, as Olshtain’s (1983) study of
English- and Russian-speaking learners of L2 Hebrew showed.
There are also other explanatory factors that may be involved. Beebe and
Takahashi (1989a) invoke accommodation theory (see Chapter 4) to suggest
that Japanese learners’ directness may reflect ‘psychological convergence’.
That is, they may attempt to converge towards a stereotypical norm of native-
speaker behaviour, but overshoot the mark, thus diverging. Such conver¬
gence can be teacher-induced, as when Japanese teachers advise their students
to be direct when in the United States, resulting in learners producing such
refusals as ‘Hell no!’. Another possibility is found in Preston’s (1989) obser¬
vation that learners may sometimes wish to appear ‘learner-like’ as this brings
communicative advantages in certain situations, a point we will take up later
when we consider impression management. Preston suggests that this may ac¬
count for the greater verbosity of advanced learners’ speech acts. It is possible
that learners vary in their style of speaking, sometimes opting to perform in
 Pragmatic aspects of learner language 183

accordance with their stereotypes of native speakers and sometimes empha¬

sizing their status as learners.
Yet another possibility, which can explain the greater verbosity observed in
advanced learners, is that learners prefer a more transparent communicative
style because it is safer. Kasper (1989) suggests that learners may be conscious
of their foreigner role and, as a result, feel a stronger need to establish, rather
than presuppose, common ground than do native speakers. This leads learn¬
ers to try to make their reasons for imposing on interlocutors as explicit as
possible when performing illocutionary acts such as requests.
Three other explanatory factors are worth mentioning. As in the case of lin¬
guistic errors, pragmatic errors may be the result of overgeneralization.
Learners may overextend a realization strategy from a situation in which it is
appropriate to one in which it is not. Also, the extent to which learners are
able to encode illocutionary acts in socially appropriate ways may depend on
whether their attention is focused on simply getting the propositional content
of their utterance across or also on its modality. Kasper (1984a) found that
when learners were under communicative pressure they tended to engage in
modality reduction,8 That is, they omitted grammatical features such as
modal verbs and adverbials associated with the expression of modal mean¬
ings like possibility and tentativeness.
This account of pragmatic development tends to assume that learners are
targeted on some set of native-speaker norms. As Blum-Kulka (1991) has
noted, such an assumption is questionable, however. Learners may prefer to
maintain their own ethnic identity or they may wish to establish a separate
identity as an L2 learner/user. In such cases, behaving in accordance with nat¬
ive-speaker norms will be perceived as inappropriate. The distinctive prag¬
matic features evident in the language of even very advanced L2 learners may
reflect not so much a failure to achieve target-language norms as the attain¬
ment of a mode of behaviour compatible with the learners’ chosen sense of
identity (see Chapter 6 for further discussion of this aspect of L2 acquisition).
This observation may apply more to sociopragmatic than pragmalinguistic
choices, for, as Thomas points out, ‘sociopragmatic decisions are social be¬
fore they are linguistic, and while foreign learners are fairly amenable to cor¬
rections they regard as linguistic, they are justifiably sensitive about having
their social... judgement called into question’ (1983: 104).

Impression management
The study of language functions such as requests, apologies, and refusals has
taken place within the framework provided by speech act theorists such as
Austin and Searle. The study of impression management, however, has been
informed by interactional sociolinguistics (Erickson and Schultz 1982;
Gumperz 1982; Scollon and Scollon 1983). This examines how speakers
achieve communicative effects by manipulating their linguistic and non-
linguistic resources.
184 The description of learner language

The choices that individual speakers make from the repertoire of signs
available to them serve as one of the ways in which they seek to form and
change the attitudes that other speakers have towards them. Speakers exploit
contextualization cues (i.e. the signals that trigger how speakers view the con¬
text they are attempting to build through interaction) to channel the listeners’
interpretations of what is being said. In particular, cues that are in some way
unexpected serve as devices for managing the listener’s impressions of a
speaker. The study of these cues has involved the qualitative analysis of social
encounters in which potential problems of communication arise because the
speakers do not share the same cues or because they attribute different values
to those cues they do share (see Gumperz 1982).
Conversations between L2 speakers and native speakers constitute a good
example of the ‘unequal’ encounters that have attracted the attention of re¬
searchers in the field of interactional sociolinguistics. Learners are likely to
find themselves in an unequal position for two reasons. First, they are unequal
in terms of their knowledge of the language they are trying to communicate
in. Second, many L2 learners find themselves trying to communicate with nat¬
ive speakers who hold positions of power and so can affect their future (for
example, in terms of employment or access to social amenities).
A simple example of how impression management works in such encoun¬
ters is provided by Ellis and Roberts (1987). An Asian applicant in Britain is
being interviewed by white managers for a job. They are discussing how the
Asian obtained a previous job:

NS: So you eventually did get a job with the assistance of a friend. Did he
recommend you to the employment centre or to the manager ...?
NNS: No—well, he brought me a form ...

The Asian’s comment that his friend assisted him by bringing him an applica¬
tion form constitutes a device for amending the interviewer’s preconception
that Asians obtain jobs illicitly by having their friends help them get round the
system. The comment constitutes a contextualization cue because it is unex¬
pected in relation to the interviewer’s preceding comment. In this case, the
Asian possesses the L2 resources to successfully redefine the context in a way
more favourable to himself.
Not all learners are so successful in impression management. Chick (1985,
cited in Wolfson 1989a) found that black students in South Africa were un¬
able to overcome the negative cultural stereotypes held by their white pro¬
fessors in discussions of their examination results despite the fact that they
were fluent L2 speakers. Chick argues that this failure arose from ‘a mismatch
of contextualization cues’. Roberts and Simonot (1987) report on a study of
minority ethnic workers in encounters with an estate agent and in more in¬
formal conversations with a researcher. They describe the interactional style
of one worker who was content to play a subservient role in the estate agent
encounter. He never offered information, he developed only those themes
 Pragmatic aspects of learner language 185

which the estate agent had explicitly sanctioned and he omitted affective re¬
sponses that might have conveyed to the estate agent that his wishes were not
satisfied. Roberts and Simonot suggest that this learner’s repeated failure to
negotiate a context in which he could participate as an equal resulted in his
failure to acquire the interactional knowledge needed to escape from his dis¬
advantaged social position. They note that this learner was typical of their
subjects.
Learners are likely to experience difficulties in identifying the right contex-
tualization cues to deal with problems that arise in interaction. Fiksdal (1989)
used the techniques developed by Erickson and Schultz (1982) to investigate
how Chinese speakers of English handled uncomfortable moments during
academic advisory sessions with native-speaker advisers. In this kind of situ¬
ation, problems arise when the learners need to reject the advice offered by
the native speaker. Fiksdal found that, unlike native-speaker students, the
Chinese students either delayed or omitted offering verbal repair of uncom¬
fortable moments and, in the case of delayed repair, showed a preference for
an implicit rather than an explicit statement. She suggests that this may reflect
their wish to avoid forcing the advisers to lose face by indicating that their ad¬
vice is not acceptable.
Sometimes learners may be able to substitute alternative contextualization
cues, by drawing creatively on their interlanguage resources. Rampton’s
(1987) study of ESL learners in a London language unit indicates one way in
which this might take place. He observed these learners using primitive in¬
terlanguage forms such as ‘me no like’ and ‘me too clever’, even though they
clearly possessed the competence to produce more target-like forms. A form-
function analysis revealed that the ‘broken English’ forms were used to per¬
form potentially face-threatening acts such as rejection/refusal and boasting.
Rampton suggests that the learners made deliberate use of ‘me’ constructions
to symbolize their cultural and social incompetence as a way of mitigating the
force of these face-threatening speech acts. By emphasizing their learner sta¬
tus, they were able to challenge the teacher or play down their attempt to
claim superiority over their peers.
To sum up, when learners participate in conversations with native speakers
and other learners—particularly if the encounters are of the unequal
kind—they need to negotiate the impression they wish to create. Frequently,
they lack knowledge of the relevant contextualization cues. One solution is to
accept the social role allocated to them—a kind of avoidance strategy. An¬
other is to substitute cues from their native language—a form of transfer. A
third solution is to make creative use of their interlanguage resources to ex¬
ploit their status as language learners. Little is currently known about how
learners’ use of contextualization cues develops over time and how they learn
to manage impressions in a manner compatible with target-language norms.
186 The description of learner language

Summary
The focus of research on interlanguage pragmatics to date has been a relat¬
ively small set of well-defined illocutionary acts. The research has made use of
spoken data collected in natural settings and also data elicited by means of
discourse completion tasks. Learners have been found to make both socio¬
pragmatic errors (as when they fail to respond to a native-speaker compli¬
ment) and pragmalinguistic errors (as when they resort to formulas and
conventionalized routines borrowed from their LI in order to express gratit¬
ude).
Requests, apologies, and refusals are three acts which have received consid¬
erable attention. The perception/understanding of requests poses few prob¬
lems for learners, but production is more problematic. Low-proficiency
learners tend to make use of a limited range of formulas, employ non-target-
like politeness strategies, and have difficulty in adjusting their strategy in ac¬
cordance with situational requirements. Advanced learners manifest know¬
ledge of a broad range of linguistic realization devices which they are able to
use with native-like appropriateness. However, their requests differ from
those of native speakers in that they tend to be verbose. In the case of L2 apol¬
ogies, both linguistic proficiency and LI transfer are also factors. Interest¬
ingly, learners differ from native speakers by both failing to say enough on
some occasions and saying too much on others. The learners’ perception of
the universality of speech act realization also influences performance. The
study of refusals by Japanese learners of English shows that they are not al¬
ways stereotypically indirect and that, when a threshold level of L2 profi¬
ciency has been achieved, pragmatic transfer is more likely to occur. As with
requests, the classroom setting may not afford learners the range of situations
needed to develop full sociolinguistic competence.
Impression management is also problematic to learners, as they may not
have access to the contextualization cues they need to construct a context
favourable to themselves. Sometimes they make no effort to do so, sometimes
they borrow cues from their LI and at other times they may even deliberately
employ interlanguage forms to exploit their identity as language learners
when performing face-threatening language functions.

Conclusion

Studies of interlanguage pragmatics have concentrated on describing the dif¬

ferences between the way in which L2 learners and native speakers perform
the same speech acts. They have also focused on the pragmatic problems that
learners experience. Less attention has been given to how learners’ pragmatic
competence develops over time. As a result, although quite a lot is now
known about how learners use an L2, very little is known about how ‘rules of
speaking’ are acquired. For this, longitudinal studies are needed.
 Pragmatic aspects of learner language 187

The studies to date suggest that three factors are of major importance in the
acquisition of pragmatic competence. The first is the level of the learners’ lin¬
guistic competence. Learners cannot construct native-speaker-type discourse
unless they possess the linguistic means to do so. Learners with limited L2
proficiency find few problems in performing the speech acts that are com¬
municatively important to them but considerable difficulty in performing
them in native-like ways. The need to achieve this may be one of the prime
factors motivating continued L2 linguistic development.9
The second factor of obvious importance is transfer. There is ample evid¬
ence that learners transfer ‘rules of speaking’ from their LI to their L2. As
Riley (1989: 247) emphasizes, cultural transfer is evident in the types of com¬
municative events that learners expect to occur in a given situation, the man¬
ner of their participation in them, the specific types of acts they perform and
the way they realize them, the way topics are nominated and developed, and
the way discourse is regulated. Yet it is important not to overstate the role of
the non-native speaker’s LI and culture. First, as we saw in the case of re¬
quests, learners have to develop a satisfactory level of linguistic competence
before transfer of complicated LI strategies and routines becomes possible.
While acknowledging the importance of transfer, we also need to recognize
that other factors are involved. Second, as Kasper and Dahl (1991) point out,
researchers of interlanguage pragmatics have tended to take a very product-
orientated view of transfer, which has been rejected in current process-
orientated accounts. Transfer needs to be viewed as a complex process, con¬
strained by other factors such as the learner’s stage of development (see Chap¬
ter 8). They comment: ‘Clarifying the concept of pragmatic transfer should
have high priority on the research agenda in IL pragmatics’ (1991: 225).
The third factor is the status of the learner. Learners do not usually parti¬
cipate in communicative events as equals—at least when their interlocutors
are native speakers. One reason for the lack of equality may be the learner’s
overall social status in the native-speaker community. This is reflected in the
discourse which has been observed to take place in ‘gate-keeping encounters’.
But learners also have a reduced status simply because they are learners. As
Harder (1980: 268) puts it, ‘the learner is a coarse and primitive character
from an interactional point of view’. Learners are, in a sense, ‘clients’, and this
status determines the kind of discourse they typically take part in and the role
they play in it. For example, adult learners in conversations with native
speakers are likely to have few opportunities to nominate topics and tend not
to compete for turns. This restricts the range of speech acts they will need to
perform. It is not yet clear what the repercussions of this are for the acquisi¬
tion of both linguistic and pragmatic competence, but there is sufficient evid¬
ence to suggest that learners may benefit from opportunities for a more equal
discourse role, such as occurs in communication with other learners.
Finally, it is should be pointed out that the study of interlanguage prag¬
matics acts in L2 acquisition has focused on the spoken medium and has paid
188 The description of learner language

little attention to writing. This is particularly the case with illocutionary acts.
In effect, therefore, although we know something about how ‘contextualized’
acts such requests, apologies, and refusals are acquired, we know little about
how learners acquire the ability to perform acts found in decontextualized,
written language. Snow’s (1987) study of the acquisition of definitions by
French/English bilingual children indicates that ‘formal definitional know¬
ledge’ constitutes an area of learning that is independent of ‘communicative
adequacy’. This suggests that the ability to perform speech acts like requests,
apologies and refusals in face-to-face interaction may be distinct from the
ability needed to perform speech acts like definitions in writing.10 Clearly, if
the study of interlanguage pragmatics is to progress it will need to examine
written as well as spoken learner language.

Notes
1 Levinson (1983) points out the term ‘speech act’ is generally used to refer
exclusively to ‘illocutionary act’.
2 Kasper and Dahl (1991: 216) specifically exclude the following aspects
from their review of research methods in interlanguage pragmatics: ‘...
conversational management, discourse organization ... sociolinguistic as¬
pects of language use such as choice of address forms’.
3 Native-speaker norms will vary according to the native-speaker variety
that is sampled. The pragmatic rules that govern the performance of spe¬
cific illocutionary acts may vary from one group of native speakers to
another.
4 It should also be noted that learners may prefer to retain the patterns of
pragmatic behaviour of their LI, i.e. refuse to conform to native-speaker
norms. This may be because they wish to maintain their own identity or
may be done for strategic reasons (see the section on ‘impression
management’ in this chapter).
5 Olshtain (1989) adds a caveat to her claim that apology strategies are uni¬
versal. She notes that it may be ‘an artefact of our data collection instru¬
ment—the seven apology-inducing situations were selected intentionally
to create contexts that were cross-culturally very similar’ (198 9:171). It re¬
mains a possibility, therefore, that there are cross-cultural differences re¬
garding when speakers apologize. It can be noted that Japanese speakers of
L2 English sometimes apologize in contexts that call for thanking—on
leaving someone’s house where they have been given dinner, for instance.
6 The semantic formulas that Beebe classifies as ‘indirect’ appear to be of two
kinds, which perhaps should be distinguished. Some of the formulas, for
example, (4) ‘Wish’, are likely to be used alongside direct refusals—for ex¬
ample, by preparing for them—while others, for example, (5) ‘Excuse, rea¬
son, or explanation’ can be used in place of a direct refusal.
 Pragmatic aspects of learner language 189

7 In some cases, learners may continue to overuse politeness markers such

as ‘please’ even when they develop alternative linguistic means. Examples
of such learners are the Japanese subjects studied by Beebe and
Takahashi.
8 Learners are not alone in ignoring modality under certain conditions.
Native speakers are also likely to forsake politeness markers if the context
does not favour the explicit indication of social positions.
9 An issue of considerable interest is whether learners first acquire linguistic
forms and then learn how to use them in appropriate ways, as variabilists
like myself propose (Ellis 1989c), or whether the linguistic forms are ac¬
quired as a product of learning how to communicate, as discourse ana¬
lysts such as Hatch (1978b) propose. It is likely that both processes are
involved, although classroom learners in particular are likely to acquire
form before function.
10 There is, of course, a substantial amount of work on the performance of
decontextualized speech acts by native speakers (see, for example,
Flowerdew’s (1992) study of definitions in science lectures).

Further Reading

The available literature on interlanguage pragmatics tends to assume a gen¬

eral familiarity with descriptive and theoretical frameworks employed in the
study of pragmatics. Information about these frameworks can be found in:
M. Coulthard, An Introduction to Discourse Analysis (Longman, 1985) and,
in a shorter form, in:
E. Hatch and M. Long, ‘Discourse analysis, what’s that?’ in D. Larsen-
Freeman (ed.), Discourse Analysis in Second Language Research (Newbury
House, 1980).
N. Wolfson, ‘Rules of speaking’ in J. Richards and R. Schmidt (eds.), Lan¬
guage and Communication (Longman, 1983), also provides a very readable
introduction.
The field of interlanguage pragmatics is somewhat disparate and there is no
comprehensive overview available.
Kasper, G. and M. Dahl, ‘Research methods in interlanguage pragmatics.’
Studies in Second Language Acquisition (1991) 13, focuses on the research
methods used to investigate L2 illocutionary acts, but in the process provides
an excellent account of the research in this area.
N. Wolfson, Perspectives: Sociolinguistics and TESOL (Newbury House,
1989), also provides a good overview.
G. Kasper, ‘Pragmatic transfer.’ Second Language Research (1992) 8, offers a
sound overview of L2 speech act research and a solid discussion of the role of
transfer.
190 The description of learner language

In general, though, the reader should turn to the original reports of the re¬
search. The following reflect the range of work undertaken and are fairly easy
to read:

1 Pragmatic errors
P. Riley, ‘Well don’t blame me! On the interpretation of pragmatic errors’ in
W. Oleksy (ed.), Contrastive Pragmatics (John Benjamins, 1989).

2 Requests
R. Scarcella, ‘On speaking politely in a second language’ in C. Yorio, K.
Perkins, and J. Schachter (eds.), On TESOL ‘79 (TESOL, 1979).

3 Apologies
E. Olshtain and A. Cohen, ‘Apology: A speech-act set’ in N. Wolfson and E.
Judd (eds.), Sociolinguistics and Second Language Acquisition (Newbury
House, 1983).

4 Refusals
L. Beebe, T. Takahashi, and R. Uliss-Weltz, ‘Pragmatic transfer in ESL re¬
fusals’ in. R. Scarcella, E. Andersen, and S. Krashen (eds.), Developing Com¬
municative Competence in a Second Language (Newbury House, 1990).

5 Impression management
B. Rampton, ‘Stylistic variation and not speaking “Normal English’” in R.
Ellis (ed.), Second Language Acquisition in Context (Prentice Hall, 1987).
Readers interested in more technical accounts might like to look at some of
the publications that have come out of the Cross-Cultural Speech Act Reali¬
zation Project, for example:
S. Blum-Kulka, J. House, and G. Kasper (eds.), Cross-cultural Pragmatics:
Requests and Apologies (Ablex, 1989).
There are also a number of relevant articles in:
S. Gass et al. (eds.), Variation in Second Language Acquisition, Volume I:
Sociolinguistic Issues (Multilingual Matters, 1989).
PART THREE
Explaining second language acquisition:
External factors
Introduction

The chapters in the previous section concentrated on the description of learn¬

er language. The chapters in the next three sections will attempt to explain
learner language by addressing three major questions:

1 How do learners learn a second language?

To answer this question it is necessary to explain how learners obtain

information about the L2 and how they process this information in order to
develop their interlanguages.

2 Why do learners vary in how fast they learn a second language?

Learners vary in the speed with which they acquire an L2. What factors ac¬
count for this inter-learner variation?

3 Why do most learners fail to achieve full target-language competencef

As we have already seen in Chapter 3, learners vary in the ultimate level of

proficiency they achieve, with most of them failing to reach target-language
competence. A full explanation of L2 acquisition must account for this
phenomenon.
To answer these questions we will need to consider three aspects of L2 ac¬
quisition: (1) the external environment, (2) the ‘black box’ (i.e. the learner’s
existing knowledge and the internal mechanisms that guide L2 acquisition),
and (3) individual learner factors, such as age and motivation.
Our search for explanations will be guided by the framework shown in the
figure below. There are three principal components in this framework: (1) so¬
cial factors/settings, (2) language processing mechanisms, and (3) individual
learner factors. These are interrelated in a number of ways. Social factors (for
example, a learner’s ethnic background) may influence individual learner fac¬
tors (for example, a learner’s motivation to learn a particular language). It is
also possible, of course, that individual learner factors (for example, a learn¬
er’s personality) may have an effect on the social conditions of learning, for
example by influencing the nature of the social settings experienced. Social
factors/settings serve as the major determinant of the input that learners re¬
ceive. For example, they influence what variety of the target language the
learners will be exposed to and also the amount of exposure. Input provides
the data for the language processing mechanisms. These are also, potentially
194 External factors

at least, influenced by individual learner factors and also by the learner’s

‘other knowledge’ (including the LI). The language processing mechanisms
account for changes in the learner’s L2 knowledge (the interlanguage system).
This knowledge is used in both the comprehension and production of utter¬
ances in the L2 (i.e. output). Output also helps to shape subsequent input, as,
for example, when what learners say influences what is said to them.
The process of L2 acquisition is very complex, and this framework simpli¬
fies it somewhat. Also, some of the relationships shown are more controver¬
sial than others. For example, as we will see in Chapter 11, not all researchers
agree that individual learner factors have a direct impact on language pro¬
cessing. However, the framework serves as a heuristic in our search for
explanations.
This section will focus on the components that lie outside the ‘black box’:
Chapter 6 will consider social factors and settings while Chapter 7 will exam¬
ine the nature of the input and interaction that learners experience and how
this influences L2 acquisition. As in subsequent sections, we will be searching
for explanations by looking for answers to all the three questions mentioned
at the beginning of this Introduction. However, it will also be necessary to re¬
view research of a descriptive nature. We need to know, for example, what
are the main characteristics of different social settings as well as how these in¬
fluence learning outcomes. Similarly, we need to understand what is the na¬
ture of the input and interaction learners experience, as well as how these
affect language processing.
The categorization of explanations in terms of external factors, internal
factors, and individual learnet differences is justified because it reflects ident¬
ifiable orientations in SLA research and because it provides a means of or¬
ganizing the information provided by research. However, it should be
emphasized that there is no single and no simple explanation of L2 acquisi¬
tion. A complete explanation will need to consider all the three components

(2)
Other knowledge
t (e.g. of LI)
(1)
social
input
I
factors/ Language -► L2 knowledge output
settings processing (IL system)
A
—-—

(3)
^ Individual
learner
factors

A framework for explaining L2 acquisition

 External factors 195

and how they interact. Social factors/settings and input/interaction cannot be

considered in total isolation from individual learner factors or language pro¬
cessing. In the two chapters in this section, therefore, while focusing attention
on external factors, we will also need to give some consideration to the other
two components shown in the framework.
.

‘f
6 Social factors and second language
acquisition

Introduction

This chapter examines the relationship between society and second language
(L2) learning. It considers the role of social factors in L2 proficiency. Learners
differ enormously in how quickly they learn an L2, in the type of proficiency
they acquire (for example, conversational ability as opposed to literacy in the
L2) and the ultimate level of proficiency they reach. In part, these differences
can be explained by reference to psychological factors such as language
aptitude, learning style and personality (see Chapter 11) but in part they are
socially determined.
This chapter will also consider the relationship between different learning
contexts and L2 proficiency. ‘Context’ here refers to the different settings in
which L2 learning can take place—whether the setting is a ‘natural’ or an
‘educational’ one, and various subdivisions of these, such as submersion and
immersion. It is, perhaps, a matter of debate as to whether such distinctions
are to be viewed as ‘social’, but the view taken here is that, in essence, they are.
Each setting can be seen as a context in which constellations of social factors
typically figure to influence learning outcomes.
Social factors have a major impact on L2 proficiency but probably do not
influence it directly. Rather, their effect is mediated by a number of variables.
One set of variables which have been found to be of major importance is
learner attitudes. Social factors help to shape learners’ attitudes which, in
turn, influence learning outcomes. Social factors also influence L2 learning in¬
directly in another way. They determine the learning opportunities which in¬
dividual learners experience. For example, the learners’ socio-economic class
and ethnic background may affect the nature and the extent of the input to
which they are exposed.
Social accounts of L2 learning have been primarily concerned with L2
proficiency. In this respect, they differ from other branches of SLA research,
which, as we saw in Part Two of this book, have focused on specific linguistic
and pragmatic properties of learner language. L2 proficiency has been con¬
ceptualized in different ways, none of which, according to Stern (1983: 356),
provides ‘a completely satisfactory expression’. It has been measured by
means of rating scales, where the assumption is that learners range from zero
198 External factors

to full native-like proficiency, standardized tests such as the TOEFL examina¬

tion, and performance on a range of different tasks designed to tap different
components of proficiency.
One model of L2 proficiency that has figured significantly in socially-
oriented SLA research is that of Cummins. In his early work, Cummins distin¬
guished two types of proficiency. Basic interpersonal communication skills
(BICS) are the skills required for oral fluency and sociolinguistic appropri¬
ateness. They are ‘basic’ in the sense that they develop naturally as a result of
exposure to a language through communication. Cognitive/academic lan¬
guage proficiency (CALP) consists of the linguistic knowledge and literacy
skills required for academic work..In his later work (for example, Cummins
1983; Cummins and Swain 1986), Cummins has proposed that language pro¬
ficiency be conceptualized along two interacting continua. One continuum
relates to the extent of the contextual support available for expressing or
receiving meaning. At one extreme, a task might require context-embedded
language where communication derives from ‘inter-personal involvement in
a shared reality’, while at the other the task might require context-reduced
language, where ‘shared reality cannot be assumed’ (Cummins and Swain
1986: 153). The other continuum concerns the extent to which a task is cog¬
nitively demanding. This reflects ‘the amount of information that must be
processed simultaneously or in close succession’ and also the extent to which
the information needed to perform the task has become automatized.
This chapter will begin with a brief discussion of learner attitudes and how
they have been studied. It will then address three broad questions:

1 To what extent do specific social factors (age, sex, social class, and ethnic
identity) affect L2 proficiency?
2 To what extent do social factors influence the learner’s choice of target lan¬
guage variety?
3 How can we characterize the different social contexts in which L2 acquisi¬
tion takes place, and what effect does the type of context have on learning
outcomes?

The chapter also provides an account of a number of social models of L2

acquisition.

Learner attitudes

Learners manifest different attitudes towards (1) the target language, (2) tar¬
get language speakers, (3) the target-language culture, (4) the social value of
learning the L2, (5) particular uses of the target language, and (6) themselves
as members of their own culture. These attitudes are likely to reflect the par¬
ticular social settings in which learners find themselves. Learner attitudes
have an impact on the level of L2 proficiency achieved by individual learners
and are themselves influenced by this success. Thus, learners with positive
 Social factors and second language acquisition 199

attitudes, who experience success, will have these attitudes reinforced. Sim-
ilarly, learners’ negative attitudes may be strengthened by lack of success. We
will also find cases of learners who begin with positive attitudes but who, for
one reason or another, experience inadequate learning opportunities, fail to
progress as they expected, and, consequently, become more negative in their
outlook.
Baker (1988) discusses the main characteristics of attitudes:

1 Attitudes are cognitive (i.e. are capable of being thought about) and af¬
fective (i.e. have feelings and emotions attached to them)—see Triandis
1971.
2 Attitudes are dimensional rather than bipolar—they vary in degree of
favourability/unfavourability.
3 Attitudes predispose a person to act in a certain way, but the relationship
between attitudes and actions is not a strong one.
4 Attitudes are learnt, not inherited or genetically endowed.
5 Attitudes tend to persist but they can be modified by experience.

Attitudes have been measured both indirectly and directly. An example of

indirect measurement is the Semantic Differential Technique. This presents
learners with a series of antonyms (for example, useful-useless; ugly-
beautiful) and asks them evaluate a given phenomenon (for example, a lan¬
guage or a speaker’s accent) on each dimension. It has been used in Matched
Guise studies (see Lambert, Hodgson, Gardner, and Fillenbaum 1960), in
which the same speaker, who is bilingual, reads a passage in two languages.
Learners are then asked to make judgements about the readers (i.e. they are
not told that it is one person). In this way, they inadvertently reveal their atti¬
tudes towards the two languages. Spolsky’s Identity Scale (Spolsky 1969) also
measures attitudes indirectly. Subjects are given a list of adjectives and asked
to say how well each adjective describes themselves, their ideal selves, people
whose native language is the same as theirs, and native speakers of the target
language.
Direct measurement of attitudes usually involves self-report question¬
naires. These often take the form of a series of statements to which the learn¬
ers respond on a five-point scale (for example, from ‘strongly agree’ to
‘strongly disagree’). Gardner and Lambert (1972) used this kind of question
extensively in their studies of attitudes among American students learning L2
French in the United States. Questions were devised to investigate learners’
reasons for learning French, their degree of anomie (dissatisfaction with their
place and role in society), their ethnocentrism, their preference for America
over France, and their attitudes towards French Americans. An example of an
item used to measure ethnocentrism is:

The worst danger to real Americanism during the last fifty years has come
from foreign ideas and agitators. Certain people who refuse to salute the
200 External factors

flag should be forced to conform to such patriotic action, or else be

imprisoned.

There is some disagreement regarding the validity and reliability of these

measures of learner attitudes. Oiler has argued that people will ‘self-flatter’ by
responding to an attitude test in a way that makes them appear more prestigi¬
ous than is the case, or else will seek to give socially desirable answers (see
Oiler 1977 and 1981) and that they are, therefore, unreliable. Gardner (1980;
1985), however, rejects Oiler’s criticisms on the grounds that it is possible to
safeguard against such responses by giving careful thought to the way the
questionnaire is designed (for example, by ensuring that there are multiple
measures of each attitude).
Learners’ attitudes may predispose them to make efforts to learn the L2 or
not to do so. In some cases, learners may be subject to conflicting attitudes.
On the one hand they wish to learn the L2 because it is seen as a way of assimi¬
lating into the majority culture, while on the other they wish to maintain their
LI as a way of affirming their own identities. A good example of this kind of
conflict can be seen in Irish people’s attitudes towards learning Gaelic. The 5
year study by the Committee on Irish Language Attitudes Research (1975)
found that there was general agreement that Gaelic was necessary for ethnic
and cultural integrity, but at the same time there was only low or lukewarm
commitment to its actual use. Edwards (1984) concluded that the Irish like
their Irish, but that they like it dead!
In general, positive attitudes towards the L2, its speakers, and its culture
can be expected to enhance learning and negative attitudes to impede learn¬
ing. This need not necessarily be so, however. Negative attitudes may have a
positive effect on L2 learning if the learners have a strong reason for learning.
Lanoue (1991) provides a fascinating account of the Sekani’s linguistic devel¬
opment in British Columbia, Canada. This remote Indian tribe rejected its
own mother tongue in favour of L2 English for even inter-ethnic communica¬
tion not because they felt positively disposed towards Anglo-Canadian cul¬
ture ((in fact, they held negative attitudes) and not because of the
socio-economic advantages of English, but because English had become a
symbol of pan-Indianism, which was seen as the only way ensuring the tribe’s
identity in the future.
Lanoue’s study testifies to the enormous complexity of the relationship be¬
tween attitudes and L2 learning. Given that much of the research has been
correlational in design (i.e. has used statistical procedures to correlate atti¬
tude measures with proficiency measures) and, therefore, does not address
cause-and-effect directly, considerable caution needs to be exercised in inter¬
preting the results.
 Social factors and second language acquisition 201

Social factors and second language acquisition

We will now consider a number of specific social factors which influence the
attitudes held by different groups of learners and which lead to different levels
of L2 proficiency. Discussion will be restricted to the four variables which
have received, the most attention in SLA research: (1) age, (2) sex, (3) social
class, and (4) ethnic identity. Although each variable will be considered separ¬
ately, it should be recognized that they interact in complex ways, thus making
it difficult to determine the precise ways in which each variable contributes to
L2 learning.

Age

The age factor is considered in some detail in Chapter 11, where psycho-
linguistic aspects of age and L2 acquisition are examined. The discussion here
is restricted to sociolinguistic aspects.
Age has received considerable attention from sociolinguists. Chambers and
Trudgill (1980), for instance, document variants of/q/ in the speech of differ¬
ent generations of speakers in Norwich (England). The younger generation
(10-19 years) used non-standard variants, while middle-aged speakers
(30-60 years) preferred the standard variant. Older speakers (70+ years)
demonstrated use of non-standard variants, although not to the same extent
as the younger generation. Chambers and Trudgill seek to explain this pat¬
tern by suggesting that younger speakers are subject to social pressures from
their peer group, while middle-aged speakers have less cohesive social net¬
works and are more influenced by mainstream societal values. In older, re¬
tired people, social pressures lessen and social networks again become
narrow.
The general pattern of social influence which Chambers and Trudgill docu¬
ment may help to explain age-related factors in L2 acquisition. Learners who
commence learning an L2 after the onset of puberty (and possibly earlier) are
unlikely to acquire a native-speaker accent, while those who begin after the
age of about 15 years are less likely to develop as much grammatical ability as
those who begin before (see Chapter 11). Preston (1989) suggests that chil¬
dren may be more prepared to share external norms because they are not sub¬
ject to peer pressure and have not formed stereotypes of their own identities.
He argues that the threat to identity in older learners occurs even in ‘short¬
term, restricted’ L2 acquisition, which may account for why many adoles¬
cents are resistant to L2 learning in foreign language settings. This explana¬
tion is not entirely convincing, however. It does not explain why adolescent
learners progress more rapidly than younger learners to begin with. Nor does
it explain why adolescents tend to do better than middle-aged learners, who
ought to outperform the younger generation given their greater acceptance of
202 External factors

prestige social norms. As Preston recognizes, a social explanation of age-

related effects in L2 acquisition is, at best, only a partial explanation.

Sex

A distinction is often made between ‘sex’ and ‘gender’. The former constitutes
a biological distinction, while the latter is a social one. A number of socio-
linguists currently prefer the term ‘gender’ because it places the emphasis on
the social construction of ‘male’ and ‘female’ (see Kramarae 1990). As Labov
notes, ‘there is little reason to think that sex is an appropriate category to ex¬
plain linguistic behavior’ (1991: 206), and so it is necessary to posit some in¬
tervening variable (i.e. the distinct roles assumed by the different sexes). The
term ‘sex’ is used here to reflect the way in which the variable has been typic¬
ally measured in SLA research (i.e. as a bipolar opposite).
Sociolinguistic research has identified two distinct and apparently contra¬
dictory principles relating to sex differentiation in native-speaker speech (see
Labov 1991:206-7):

1 In stable sociolinguistic stratification, men use a higher frequency of non¬

standard forms than women.
2 In the majority of linguistic changes, women use a higher frequency of the
incoming forms than men.

Women, therefore, nearly always outstrip males in the standardness of their

speech and use of prestige forms, and yet they also tend to be in the forefront
of linguistic change. This can be explained by positing that women are more
sensitive to new forms and more likely to incorporate them into their speech,
but, when they become aware of the change, they are inclined to reject them.
Men, on the other hand, may be less sensitive to new forms but once they have
started to use them are less likely to reject them, perhaps because they are less
likely to notice them. Both principles suggest that women might be better at
L2 learning than men; they are likely to be more open to new linguistic forms
in the L2 input and they will be more likely to rid themselves of interlanguage
forms that deviate from target-language norms.
These predictions based on sociolinguistic theory are borne out by several
studies. Female learners generally do better than male. Burstall (1975), for ex¬
ample, investigated sex differentiation in her longitudfriaTsTu3y“of some
6,000 children beginning L2 French at eight years old in English primary
schools. She reports that the girls scored significantly higher than the boys on
all tests measuring achievement in French throughout the period of the study.
Boyle (1987) reports on a study of 490 (257 male and 233 female) Chinese
university students in Flong Kong. The female students achieved higher over¬
all means on ten tests of general L2 English proficiency and in many cases the
differences were significant.
 Social factors and second language acquisition 203

Other studies, however, have produced conflicting results. Boyle (1987) re¬
ports that the male students in his study performed better on two tests of lis¬
tening vocabulary, lending support to LI research that boys are superior in
this particular aspect of language proficiency. However, Nyikos (1990) re¬
ports that women outperformed men in a German vocabulary memorization
task. Also, some studies have reported no or few differences between males
and females. Bacon (1992), for instance, found no difference between the
sexes in two authentic listening tasks.
A number of studies suggest that females have more positive attitudes to
learning an L2 than males. Burstall, in the study referred to above, notes that
low-achieving boys tended to drop French to a significantly greater extent
than low-achieving girls from the age of 13 onwards. Furthermore, the girls
displayed consistently more favourable attitudes towards learning French
than did the boys. Gardner and Lambert (1972) also report that female learn¬
ers of L2 French in Canada were more motivated than male learners and also
had more positive attitudes towards speakers of the target language. Spolsky
(1989) found that girls learning L2 Hebrew in Israel (a majority language set¬
ting) demonstrated more favourable attitudes to Hebrew, Israel, and Israelis
than boys. One study (Ludwig 1983) found that male learners were more in-
strumentally motivated (i.e. more motivated to learn the L2 for purely func¬
tional reasons), while another (Bacon and Finnemann 1992) reported that
female learners of L2 Spanish at university level had the stronger instrumental
motivation.
Other studies suggest that women tackle the task of learning an L2 differ¬
ently from men. Gass and Varonis’ (1986) research on sex differences in inter¬
actions involving learners concluded that men use the opportunities to
interact to produce more output, whereas women use it to obtain more input.
However, Pica et al. (1991) failed to find much evidence to support sex differ¬
ences in interactions involving adult male and female Japanese learners of L2
English. Clearer evidence for sex differences comes from self-report studies of
the strategies learners use. Bacon (1992) found that men reported using trans¬
lation strategies more than women, while the women reported monitoring
their comprehension more. Bacon and Finnemann (1992) found that women
reported greater use of a ‘private/non-oral mode’ in language learning than
men (for example, they relied on their LI to make the L2 meaningful, re¬
hearsing in their heads before they spoke and guessing at what might be going
on). It is not clear, however whether such questionnaires reflect the actual
practices learners follow or their ideas about what constitutes socially appro¬
priate answers.
It is not easy to find clear-cut explanations for these results, nor, perhaps,
should any be attempted at the present time. The explanations that follow are
speculative in nature. One obvious explanation for females’ greater success in
L2 learning in classroom settings is that they generally have more positive at¬
titudes. This, in turn, may reflect their employment expectations. Girls may
204 External factors

perceive a foreign language as having significant vocational value for them,

whereas boys do not. These beliefs may derive from the students’ parents.
However, although this explanation is convincing for ‘foreign’ languages
learnt in Europe or the United States, it is less clearly applicable for ‘second’
language learning in situations such as the United States or Hong Kong. It is
possible that general differences associated with male and female ‘culture’ are
also involved. Maltz and Borker (1982) suggest that girls are more likely to
stress co-operation and that they learn to deal sensitively with relationships,
whereas boys emphasize establishing and maintaining hierarchical relations
and asserting their identity. The femaie ‘culture’ seems to lend itself more
readily to dealing with the inherent threat imposed to identity by L2 learning.
Another possible explanation is that females benefit from more and better in¬
put as a result of their superior listening comprehension skills, which some,
but not all, studies have reported. There is some evidence to suggest that fe¬
males are more sensitive to input (for example, the Gass and Varonis study).
Eisenstein (1982) found that females consistently and significantly out¬
performed males in discriminating among different American English ac¬
cents. Female learners may also be more active strategy users.
A number of puzzles remain. Why, for instance, do males have a wider lis¬
tening vocabulary? Boyle offers a number of explanations, none of which are
very convincing (for example, he cites Carroll (1969), who suggests that boys
in a male culture interact over a wider range of subjects and thus develop a
broader recognition vocabulary). It is particularly puzzling given the evidence
that women are better at vocabulary learning. Perhaps it is necessary to con¬
sider sex differences in language learning in relation to both what learners
know (where women are generally superior) and how they use this know¬
ledge under varying performance conditions (where men may sometimes
prove superior).
Sex (or gender) is, of course, likely to interact with other variables in de¬
termining L2 proficiency. It will not always be the case, therefore, that fe¬
males outperform males. Asian men in Britain generally attain higher levels of
proficiency in L2 English than do Asian women for the simple reason that
their jobs bring them into contact with the majority English-speaking group,
while women are often ‘enclosed’ in the home. Sex interacts with such factors
as age, ethnicity, and, in particular, social class.

Social class

An individual’s social class is typically determined by means of a composite

measure that takes account of income, level of education and occupation. It is
customary to distinguish four groups: lower class, working class, lower mid¬
dle class, and upper middle class.
As Preston (1989: 117) points out, there is a clear parallel between socio-
linguistic phenomena associated with social class and language change and
 Social factors and second language acquisition 205

interlanguage development. Preston compares a number of sociolinguistic

and interlanguage features. Hypercorrection (for example, the overextension
of a feature like /h/ to words such as ‘hour’), which occurs when lower-
middle-class speakers seek to incorporate a prestige feature into their careful
speech, parallels overgeneralization in learner language. Hypo correction (the
retention of an old norm which has covert prestige in the speech of the work¬
ing class) is like negative transfer. An indicator (defined by Preston as ‘a form
not involved in change’) is similar to fossilization (the persistence of a non¬
standard form in interlanguage). Change from above can be compared to
monitoring (Krashen 1981—see Chapter 9), in so far as both involve con¬
scious attention to linguistic form. However, as Preston recognizes, although
these processes seem very similar, it does not follow that they are motivated
by the same social factors. Indeed, there is no reason to expect any relation¬
ship between social class and overgeneralization, negative transfer, etc. The
similarity may rest more in the psychological processes which underlie both
linguistic change and L2 learning.
There is evidence of a relationship between social class and L2 achieve¬
ment, however. Burstall (1975; 1979) found that for both male and female
primary and secondary school learners of L2 French there was a strong cor¬
relation between socio-economic status and achievement. Children from
middle-class homes regularly outperformed those from lower- and working-
class homes. There were also class-related differences in the learners’ atti¬
tudes. Working-class children tended to drop French after their second year
in secondary school, while middle-class children were likely to continue.
Olshtain, Shohamy, Kemp, and Chatow (1990) investigated the levels of pro¬
ficiency in L2 English reached by 196 grade 7 learners in Israel. The learners
were divided into an ‘advantaged’ and a ‘disadvantaged’ group on the basis of
socio-economic status. Olshtain et al. found that the two groups differed sig¬
nificantly in LI (Hebrew) cognitive academic level proficiency (CALP) and
that a number of measures of this correlated significantly with L2 English
achievement. One interpretation of this result was that the ‘advantaged’ chil¬
dren were better at learning English in a classroom setting because they had a
more developed LI CALP. Interestingly, variance in the ‘advantaged’ group
was not attributable to differences in self-reported attitudes and motivation,
whereas in the ‘disadvantaged’ group it was. Overall, though, LI CALP ex¬
plained much more of the variance in L2 achievement than did motivation
and attitudes. Finally, Skehan (1990) also reports moderate correlations be¬
tween the family background of 23 secondary school children in Bristol and
both language learning aptitude and foreign language achievement in French
and German, with middle-class children again outperforming lower-class.
Skehan suggests that these relationships may reflect the learners’ underlying
ability to deal with context-disembedded language, thus bearing out Olshtain
and her colleagues’ main conclusion.
206 External factors

All these studies examined L2 achievement in foreign language classrooms.

Their results mirror the general finding that children from lower socio¬
economic groups are less successful educationally than those from higher
groups. Another study, however, suggests that the disadvantage in language
learning shown by lower-status groups is not inevitable. Holobrow, Genesee,
and Lambert (1991) report on a study of partial immersion involving kinder¬
garten and grade 1 pupils in Cincinnati (USA). They found no difference in
either French listening comprehension or oral production in children from
different socio-economic and ethnic groups:

... the working-class and black students were able to benefit from the se¬
cond language experience as much as middle-class and white students. In
other words, the disadvantaged students were not disadvantaged when it
came to second language learning (1991: 194).

One possible reason for this is that the immersion programme placed greater
emphasis on BICS. The researchers suggest that ‘the development of oral/
aural interpersonal communication skills in a second language do not appear
to be dependent on individual differences of a cognitive, linguistic and ...
social nature.’ In other words, where BICS are concerned, social differences in
learners have no effect.
There have been few studies investigating social class and L2 learning. The
results to date suggest that middle-class children achieve higher levels of L2
proficiency and more positive attitudes than working-class children when the
programme emphasizes formal language learning. This may be because they
are better able to deal with decontextualized language. However, when the
programme emphasizes communicative language skills, the social class of the
learners has no effect.
It is important to recognize, however, that it is not socio-economic class per
se that produces these effects, but rather the experiences of the world which
members of the different social classes are likely to have. Heath’s (1983) eth¬
nographic study of children in two neighbouring communities demonstrates
how contrasting life experiences eventually lead to different levels of school
achievement. Heath shows how the types of language use working class black
children experience at home differs from that found in the classroom and thus
puts them at an immediate disadvantage in comparison to white children,
where there is greater congruity between language use at home and at school.
For example, ‘no one lifts labels and features out of their contexts’ and ‘no
one requests repetitions’ (1983: 353) in the black children’s homes, whereas
this is common in the classroom. Conversely, the black children are encour¬
aged to create ‘highly imaginative stories’ at home, but not in the classroom.
Heath’s detailed analysis supports Labov’s (1969) earlier contention that it is
‘difference’ and not ‘deficit’ that is at the root of many of the language prob¬
lems black children face at school. Although neither Labov nor Heath were
concerned with L2 learning, their arguments are of obvious relevance.
 Social factors and second language acquisition 207

Ethnic identity

Gudykunst and Ting-Toomey (1990) inform us that ‘ethnicity is a slippery

concept’. This is partly because there is tension between ‘objective’ and ‘sub¬
jective’ definitions of the term. Objective definitions predominated in early
anthropological studies, in which researchers imposed external categor¬
izations on their subjects. Subjective definitions see ethnicity as a process
whereby individuals use labels to define themselves in communication with
others and are now generally favoured. Gudykunst and Ting-Toomey point
out that both self- and other categorizations of ethnic identity may influence
the way language is used in inter-ethnic communication.
There is a general consensus that ethnic identity can exert a profound influ¬
ence on L2 learning. This influence can take three possible forms, correspond¬
ing to normative, socio-psychological, and socio-structural views of the
relationship. As these perspectives (and the theories that derive from them)
are considered in depth later (see page 230), we will focus here on representat¬
ive samples of the kinds of research to which each has given rise.
Research based on a normative view of the relationship between ethnic
identity and L2 learning seeks to establish to what extent membership of a
particular ethnic group affects L2 achievement. A key concept here is that of
the ‘distance’ between the cultures of the native and target languages, the idea
being that the more distant the two cultures are, the more difficult L2 learning
is and, therefore, the lower the achievement levels.
Svanes (1988) investigated the acquisition of L2 Norwegian by three ethnic
groups in Norway. One group (the ‘near’ group) consisted of learners from
Europe and America who shared a common ‘western’ culture. The second
group (the ‘intermediate’ group) consisted of learners from the Middle East
and Africa, all of whom had contact with western culture. The third group
(the ‘distant’ group) contained students from Asian countries (for example,
India and Vietnam). Svanes found a clear relationship between cultural dis¬
tance and L2 achievement, measured by an examination that tested a wide
variety of knowledge and skills. The Western students had the best grades, the
Middle Eastern and African students the next best, and the Asians the poorest
results. It should be noted, however, that there is no way of knowing whether
the difference in the grades obtained by the three groups was a reflection of
cultural distance or linguistic difference (see Chapter 8 for a discussion of the
role of linguistic difference).
A socio-psychological view of the relationship between ethnic identity and
L2 proficiency emphasizes the role of attitudes. The attitudes that learners
hold towards the learning of a particular L2 reflect the intersection of their
views about their own ethnic identity and those about the target-language
culture. These views will influence both L2 and LI learning, as shown in
Table 6.1.
208 External factors

Lambert (1974) distinguishes additive and subtractive bilingualism. In the

former learners maintain their LI, adding the L2 to their linguistic repertoire.
In such cases, learners may become balanced bilinguals. This is likely to occur
when learners have a positive view of their own ethnic identity and of the tar¬
get-language culture. In the case of subtractive bilingualism learners replace
their LI with the L2, failing to develop full competence in their mother tongue
or, in some cases, actually losing competence that has already been acquired.
This arises when learners have a low estimation of their own ethnic identity
and wish to assimilate into the target-language culture. When learners have
negative attitudes towards both their own culture and that of the target lan¬
guage, semilingualism may result. That is, the learners may fail to develop full
proficiency in either language. It should be noted, however, that semilingual-
ism (so defined) is a controversial notion, as it runs the danger of depicting as
deficit what is in fact only difference, and that the native LI language com¬
petencies of poor minority language children may be fully functional in the
out-of-school contexts in which they are used. This point has already been
made with reference to Heath’s work on black and white children and is fur¬
ther discussed in Edelsky et al. 1983. Monolingualism (i.e. failure to acquire
the L2) is associated with a strong ethnic identity and negative attitudes to¬
wards the target-language culture. Of course, the relationship between ethnic
attitudes and the different kinds of ‘lingualism’ is not as absolute as shown in
Table 6.1. It does not follow, for instance, that learners with negative atti¬
tudes towards the target-language culture will invariably fail to learn the L2,
as other factors (such as instrumental need) can affect the outcomes.

Attitudes towards
Native culture Target culture

Additive bilingualism + +
Subtractive bilingualism +
Semilingualism
Monolingualism +

Key: + = positive attitudes

- = negative attitudes

Table 6.1: Attitudes and L2 learning

The role of attitudes in L2 learning has been extensively researched by

Lambert and Gardner and their associates (for example, Gardner and Lam¬
bert 1972; Gardner 1985), primarily in Canada but also in other settings (for
example, the United States and the Philippines). The theoretical framework
which has informed these studies is described in the final section of this chap¬
ter, so here we will focus on some of the research it has led to.
A number of studies have supported Gardner and Lambert’s original claim
that a socially based motivation involving a ‘willingness to be valued mem-
 Social factors and second language acquisition 209

bers of the (second) language community’ (Gardner and Lambert 1959: 271)
results in high levels of L2 proficiency (see Gardner and Clement 1990).
Learners’ attitudes also affect language attrition. Gardner, Lalonde, and
McPherson (1985) found that learners of L2 French with favourable attitudes
showed little decline, while those with less favourable ones showed significant
loss in self-rated proficiency six months after an intensive course. However,
other studies suggest that the relationship between positive attitudes and L2
proficiency is less clear-cut. In some cases no significant relationship has been
found and in others there have been negative correlations. For example, Oiler
(1977) found that Chinese students with high levels of L2 English rated
Americans lower on traits such as cleverness and happiness than did those
with lower levels. Svanes, in the study referred to on page 207, found that the
Asian group, which had the lowest level of achievement, displayed the most
positive attitudes towards Norwegians. In this study, too, there was a negat¬
ive correlation between attitudes and language proficiency. Svanes suggests
that ‘for groups of adult students living in a foreign country, it is more import¬
ant to have a balanced and critical attitude to the host people than to admire it
uncritically’ (1988: 365-6). It is also possible, of course, that learners of a
non-international language like Norwegian may respond positively to Nor¬
wegian people and their culture while showing little interest in their language.
The setting is likely to affect the nature of the relationship. There is an obvi¬
ous difference between Mexicans learning English in California and native
speakers of English learning French in Canada, for example. Whereas the for¬
mer may well feel their ethnic identity is under threat from the majority cul¬
ture, the latter are likely to feel secure as members of the majority culture.
Also, as Okamura-Bichard’s (1985) study of Japanese children temporarily
residing in the United States showed, socially determined attitudes interact
with learners’ personal views. Okamura-Bichard argues that the ‘personal
translation of social factors is ... critical in motivating individual learners to
make efforts in their learning attempts’ (1985: 85). She suggests that what she
calls the ‘happiness’ factor may be more important than interest in or atti¬
tudes towards the target language when the learner is a young child. It is not
surprising, perhaps, that conflicting results have been obtained.
A socio-structural view of the relationship between attitudes and L2 learn¬
ing is evident in work which has examined the effect that ethnic identity has
on the interactions between members of different ethnic groups. This view
has been explored within the general theoretical framework of interpersonal
accommodation discussed in Chapter 4. According to ethnolinguistic identity
theory (Giles and Johnson 1981), members of an in-group may or may not
adopt positive linguistic distinctiveness strategies when communicating with
members of an out-group. Giles and Ryan (1982) suggest that speakers evalu¬
ate a situation and then decide whether to adopt status or solidarity, and per¬
son-centred or group-centred strategies. In situations where people
emphasize solidarity with their own in-group, linguistic divergence from the
210 External factors

out-group is likely, whereas in situations where they are more concerned with
status and are person-centred, convergence is likely. Language attitudes play
an important role in this model, too, but whereas Gardner and Lambert see
attitudes as affecting learning outcomes via motivation, Giles and his associ¬
ates see them as influencing learning via the nature of the inter-ethnic com¬
munication that takes place. Successful L2 learning is held to occur when
learners engage in frequent and long-term convergence.
There have been no longitudinal studies of learners who tend towards di¬
vergence or convergence. Sato (1981) showed that ethnicity affects the com¬
munication styles found in ESL classroom discourse. Asian learners
participated in fewer self-selected and teacher-allocated turns than did non-
Asian learners. She suggests that this reflected the Asian learners’ reluctance
to accommodate to American ways of speaking in the classroom. Sato did not
investigate the effect of these different patterns of interaction on L2 learning,
although she speculates that the Asians might be disadvantaged.
Giles and Byrne (1982) have drawn on ethnolinguistic identity theory to
propose a theory of L2 learning known as the Inter-group Theory. This is dis¬
cussed on page 234.

Summary

In this section we have examined in what ways specific social factors affect L2
learning. With regard to age, it has been found that younger learners are gen¬
erally more successful than older learners, possibly because their identity is
less threatened by target-language norms. In the case of sex, mixed results
have been obtained, but female learners generally outperform male learners
in language classroom settings and also display more positive attitudes. Male
learners do better in listening vocabulary, however. The effects of social class
may depend crucially on the setting; in language classrooms that emphasize
formal language learning, working-class children are often less successful
than middle-class children, whereas there is some evidence to suggest that in
immersion settings they do just as well. The central factor, and the one that
has attracted the most attention, is ethnic identity. A normative view emphas¬
izes the effect of ‘cultural distance’ on L2 learning; learners who are close to
the target-language culture are likely to outperform those who are more dis¬
tant. A socio-psychological model emphasizes the role of attitudes. The rela¬
tionship between attitudes and L2 learning is almost certainly bi-directional
and dynamic, and is likely to vary according to setting. In general, learners
with positive attitudes towards their own ethnic identity and towards the tar¬
get culture can be expected to develop a strong motivation and high levels of
L2 proficiency while also maintaining their own LI. Successful L2 learning is
also possible, however, in learners with non-integrative attitudes towards the
target culture. Attitudes based on learners’ sense of ethnic identity can also af¬
fect the nature of the interactions in which learners participate. Learners who
 Social factors and second language acquisition 211

are status- and person-centred are more likely to converge on L2 norms and
therefore more likely to be successful learners than those whose solidarity
with their own in-group encourages divergence.
It is clear that the relationships between these four factors and L2 learning
is extremely complex. It should be recognized that it is not age, sex, social
class, or ethnic identity that determine L2 proficiency, but rather the social
conditions and attitudes associated with these variables. Also, the factors in¬
teract among themselves, and their effect on learning depends to a large ex¬
tent on the setting. Any conclusions, therefore, need to be cautious.

Learners’ choice of target language variety

One way in which social factors of the kind considered in the previous section
impact on L2 learning is in terms of the influence they exert on the learners’
choice of target variety. Much of SLA research has been predicated on the as¬
sumption that learners are targeted on the standard dialect of the L2. Such an
assumption, however, is not warranted in the case of many learners. As Beebe
(1985: 404) observes, learners may be ‘active participants in choosing the tar¬
get-language models they prefer’. They do not just learn ‘a language’ but
rather ‘adopt a variety or varieties of that language’. Some deviations from
standard English, therefore, may not be ‘errors’, but may simply reflect the
dialect which the learner has targeted. For example, learners who display
copula deletion may do so because they are targeted on Black English Verna¬
cular, in which this feature is the norm.
The choice of reference group depends crucially on the social context and
how this shapes learners’ attitudes towards the different varieties with which
they come into contact. L2 learners in settings where the target language
serves as a medium of communication appear to be sensitive to differences in
the dialects to which they are exposed. Eisenstein (1982) found adult ESL
learners in the New York metropolitan area were able to recognize dialect dif¬
ferences early on in the learning process, but that their ability to categorize the
specific varieties developed more slowly. The same study also found that dia¬
lect sensitivity and attitude formation develop in parallel and that the ad¬
vanced-level learners had assimilated the attitudes associated with native
speakers of English. Goldstein (1987) also found evidence of learners’aware¬
ness of the sociolinguistic rules of dialect use. One of her learners commented:

I think we should learn to talk different ways ... when you outside, they
talk different ways, so to the right people you talk the right English (1987:
429).

Learners also seem to find some dialects more intelligible than others.
Eisenstein and Verdi (1985) report that 113 adult working-class ESL learners
of mixed proficiency levels and with different LI backgrounds found black
American more difficult to understand than either the regional standard or
212 External factors

New Yorkese (a non-standard English spoken in metropolitan New York).

This was considered ‘surprising’ because the learners had considerable expos¬
ure to black English and, in some cases, actually exhibited black features in
their interlanguage production. One explanation for this may lie in the learn¬
ers’ recognition of the low prestige of black American English. Eisenstein and
Verdi found that the learners’ attitudes to the dialects mirrored their ability to
comprehend them.
Other studies have set out to investigate which social variables influence a
learner’s ‘preference model’. Beebe (1985) reviews the sociolinguistic literat¬
ure dealing (mainly) with native speakers’ preference models, and concludes
that both ‘unmarked’ and ‘marked’ choices are evident. The unmarked
choices reflect what is ‘frequent, basic or expected’. Unmarked choices can be
seen in learners’ preferences for the variety used by peers over teachers, peers
over parents, own social group over other social group, friends over non¬
friends, high-contact over low-contact group, and high-prestige over low-
prestige group. Marked choices constitute a ‘secondary type of preference’.
Beebe emphasizes that they are just as principled and systematic as the
unmarked choices. To explain the choice an individual learner makes Beebe
suggests that learners decide whether to give precedence to solidarity or sta¬
tus. Learners with a solidarity orientation are likely to manifest unmarked
choices, while those with a prestige orientation will make a marked choice.
There have been few empirical studies of L2 learners’ (as opposed to native
speakers’) preferences, however. An exception is Goldstein’s (1987) study of
the preferred variety of 28 advanced Hispanic ESL learners from urban high
schools in the New York metropolitan area, and what factors could account
for this preference. Two linguistic variables associated with black English
were examined; negative concord and distributive ‘be’. Samples of the learn¬
ers’ speech were collected by means of an interview and a role-play. Goldstein
found significant correlations between the amount of reported contact with
black Americans and presence of both variables in the learners’ speech. How¬
ever, reported identification with black Americans did not prove to be signi¬
ficantly related to the two variables, although Goldstein queries whether the
measures of identification she obtained were valid. Goldstein concludes by
suggesting a number of other variables that may affect learners’ preference
model: covert prestige of the target-language group, the status of the target-
language group vis-a-vis one’s own or one’s desired status, the difficulty or
ease of establishing and maintaining relationships with members of the tar-
get-language group, the attitudes of one’s own ethnic group to the target-
language group and vice versa, and the instrumental value of using the target
language. However, little is currently known about the effect these variables
have.
In language settings where the L2 serves an as official language (for ex¬
ample, India or Nigeria), the reference group for many learners is not a native
speaker but rather educated users of the L2 in the learners’ own country. This
 Social factors and second language acquisition 213

is one of the main reasons for the emergence of local standards. As Kachru
points out, ‘the concept “native speaker” is not always a valid yardstick for
the global uses of English’ (1986: 17). He points out, however, that there may
be ‘schizophrenia about the perceived model and actual behaviour’. Thus,
learners may claim to be learning standard British or American English, both
of which are seen as prestige varieties, while, in fact, they are learning a local
standard. Concern over the development of local standards has been ex¬
pressed by some (see, for example, Prator’s (1968) critique of ‘the British her¬
esy’) on the grounds that they may become unintelligible to speakers of other
varieties, but such worries have probably been overstated and, in any case,
little can be done to change the situation, as in most cases learners have no op¬
portunity for contact with speakers of what Kachru calls the ‘inner circle’ (i.e.
countries where English is learnt as a native language).
In foreign-language settings the preference model is nearly always a stand¬
ard variety of the inner circle. Whether the British or American variety is pre¬
ferred will largely reflect areas of geographical influence (for example, British
in Europe and American in the Philippines and Japan). In a sense, though, it is
doubtful whether learners in these settings seriously aspire to native-speaker
levels. To do so would constitute a threat to the learners’ own ethnic identities
and also might not be favourably received by native speakers. Janicki (1985)
comments:

It has been noticed that non-natives are likely to face social consequences
when their linguistic behaviour complies with sociolinguistic rules saved
(by some norm) for the natives. Examples are the usage of obscenities,
slang expressions, or very formal pronunciation. It seems that there exists a
set of as yet unidentified norms which proscribe the use of some forms on
the part of the non-native speaker.

Such constraints seem to operate more stringently when learners are ident¬
ified as belonging to a foreign-language setting than when they are seen as
part of a setting where the target language is widely spoken, although even
here, resistance to learners adopting native-speaker norms has been observed.
(See Eisenstein 1983 for a review of native-speaker reactions to non-native
speech.) Preston (1981) suggests that an appropriate model for the L2 learner
is that of ‘competent bilingual’ rather than a native-speaker model. This may
well be the implicit model of many learners in foreign-language settings.
Figure 6.1 schematizes the factors involved in learners’ choice of language
variety. Social context and learner attitudes interact with each other to deter¬
mine the learners’ preference model. This influences the T2 input learners are
exposed to and, thereby, the linguistic characteristics of the developing
interlanguage system.
214 External factors

Learners’
attitudes
Preference Developing
Input
model IL system

Social
context _

Figure 6.1: The learner’s preference model and developing interlanguage system

The social contexts of L2 learning

This discussion of the learner’s reference group has inevitably involved a con¬
sideration of the type of context or setting in which learning takes place. We
have seen that learners’ choices are constrained by the contexts they find
themselves in. In this section we adopt a macro-social approach by identi¬
fying a number of general contexts of learning that have been shaped by so¬
cial, economic, and political forces of various kinds. We consider to what
extent successful language learning is likely in each. As Tollefson (1991) has
noted, most models of L2 acquisition, including those examined in the final
section of this chapter, are based on a neo-classical view of language learning,
according to which individual learners make choices by weighing up the per¬
sonal benefits and costs of learning the language. There is a need, however, to
also consider how structural factors shape these choices.

Natural versus educational settings

A general distinction can be made between ‘natural’ and ‘educational’ set¬

tings. The former arise in the course of the learners’ contact with other speak¬
ers of the L2 in a variety of situations—in the workplace, at home, through
the media, at international conferences, in business meetings, etc. There will
be some learners who experience the L2 entirely in natural settings and others
whose only contact with it is in educational settings. However, many learners
will be exposed to the L2 in both natural and educational settings.
4 general assumption is that the learning that takes place in natural and
educational settings is very different in nature. In natural settings informal
learning occurs. That is, learning is considered to result from direct participa¬
tion and observation without any articulation of the underlying principles or
rules (see Scribner and Cole 1973). Also, there is an emphasis on the social
significance of what is being learnt rather than on the mastery of subject mat¬
ter. In contrast, formal learning is held to take place through conscious atten¬
tion to rules and principles and greater emphasis is placed on mastery of
‘subject matter’ treated as a decontextualized body of knowledge. Krashen
(1976) distinguishes two ways in which knowledge of an L2 can be
developed; acquisition and learning (this distinction is discussed more fully
in Chapter 9). The former takes place subconsciously as a result of under¬
standing what has been said (or written) in communication, and clearly
 Social factors and second language acquisition 215

corresponds to informal learning. The latter, which involves conscious

attention to linguistic forms, corresponds to formal learning.
The correlation between informal learning and natural settings on the one
hand, and formal learning and educational settings on the other, is at best
only a crude one (see Krashen 1976). Learners in natural settings often resort
to conscious learning and may deliberately seek out opportunities to practise
specific linguistic items they have studied, as Lennon’s (1989) study of ad¬
vanced German learners of English in Britain demonstrates. Conversely,
learners in classrooms may not be required to treat the language as ‘subject
matter’, but instead be given opportunities for acquisition. As d’Anglejan
(1978) has noted, the correlation between educational settings and formal
language learning depends on the pedagogic approach. In the case of the ‘tra¬
ditional’ approach (characterized by the explicit teaching of the language),
there may be few opportunities for informal learning. But in the case of ‘in¬
novative’ approaches (as illustrated by the immersion bilingual programmes
to be discussed below), informal learning is not only possible but is actively
encouraged. There is, therefore, no necessary connection between setting and
type of learning. However, the social conditions that prevail in natural and
educational contexts may predispose learners to engage in informal or formal
learning strategies. For example, as McNamara (1973) has observed, it is rare
that teachers and learners participate in the spontaneous, meaningful inter¬
changes that are characteristic of ‘street’ learning.
Another common assumption is that natural settings lead to higher levels
of L2 proficiency than educational settings. Schinke-Llano (1990: 216), for
instance, claims that ‘second’ language acquisition results in native-like use of
the target language, while ‘foreign’ language acquisition does not. This as¬
sumption is also evident in the ‘year-abroad’ built into university level foreign
language education in many European countries, and the growing popularity
of ‘home-stay’ programmes among Japanese learners. The aim of these is to
provide foreign language learners with opportunities for informal learning,
so that they can reach higher levels of oral proficiency.
There is some support for this position. D’Anglejan (1978) reports that
Canadian civil servants freed from their jobs for as long as a year to improve
their L2 proficiency in intensive language classes did not generally become
fluent in the L2, despite a strong motivation to learn. D’Anglejan suggests
that one reason for this was the absence of any opportunities for contact with
native speakers. In contrast, Vietnamese immigrants in California who were
placed in occupational settings after a short training programme proved
highly successful. Fathman (1978) found that 12- to 14-year-old ESL learners
in the United States achieved a higher level of oral proficiency than EFL learn¬
ers of similar ages in Germany, and also displayed greater strategic compet¬
ence. The ESL learners’ speech was rated higher in fluency than in
grammaticality, while the opposite was true of the EFL learners. Interest¬
ingly, however, the ESL learners were much less certain about their ability to
216 External factors

speak English than the EFL learners, perhaps because they were using a differ¬
ent yardstick to measure themselves against. Gass (1987) conducted a com¬
parative study of ESL/EFL learners and also of Italian as a second language
(ISL) and as a foreign language (IFL). Focusing on sentence interpretation
strategies, she found no difference between the EFF/ESF groups, but a signi¬
ficant difference in the case of the IFF/ISF. Gass notes that sentence inter¬
pretation in English is unproblematic because it rests primarily on one type of
cue (word order), whereas in Italian it is more problematic because several
cue types (word order, inflectional, and pragmatic) are utilized. Thus, ‘for¬
eign’ and ‘second’ language learners will manifest no differences where relat¬
ively simple learning targets are involved, but will differ when the targets are
more complex. Complex rules cannot easily be taught, and classrooms do not
offer sufficient input for them to be learnt naturally.
It is by no means certain, however, that naturalistic settings lead to high
levels of proficiency—even oral proficiency. Fathman (1978) observes that
there was considerable variation in the levels achieved by the ESF learners in
her study, suggesting that some at least were not so successful. Schinke-
Flano’s claim that ‘second’ language learning is characterized by native-
speaker levels of ability is very doubtful. Gass observes that ‘most learners, be
it classroom or non-classroom learners, do not attain complete mastery of an
F2’ (1990: 37), and queries whether native-language proficiency is ever pos¬
sible for adults—a point taken up in Chapter 11 when the role of age in F2 ac¬
quisition is examined. Fongitudinal studies, such as those reported by
Schumann (1978b), Schmidt (1983), Klein and Dittmar (1979), and Meisel
(1983) also show that learners in natural settings often fall far short of native-
language proficiency.
Nor is it possible to conclude that learning in natural settings results in
higher levels of grammatical competence than learning in educational set¬
tings. Indeed, as we shall see in Chapter 14, there is growing evidence to sug¬
gest that learners who receive formal instruction become more grammatically
accurate than those who do not.
The notions of ‘natural’ and ‘educational’ settings are inevitably somewhat
crude. The differences within each setting regarding both quality and quant¬
ity of learning opportunities are likely to exceed the differences between
them. As a result, comparisons of the learning outcomes associated with each
setting are of doubtful value. A more useful approach, perhaps, is to examine
what factors within each setting are important for successful F2 learning. We
will do this by identifying and discussing more specific learning contexts.

Natural contexts

Following Judd (1978), three broad types of context in natural F2 learning

settings can be identified. The type that has been most thoroughly researched
is found in situations where the target language serves as the native language
 Social factors and second language acquisition 217

(or one of the native languages) of the country—as is the case for L2 learners
of English in the United States or Canada or L2 learners of French in France or
Belgium. The second type is found in the decolonized countries of Africa and
Asia, where the F2 functions as an official language. In such contexts it is not
learnt as a mother tongue by more than a few people. The third type occurs
when the F2 is used for interpersonal communications (usually of fairly spe¬
cific kinds) in countries where it is neither learnt as a mother tongue nor used
as an official language (for example, the use of F2 English for business com¬
munication in Japan). In all three types of context, situations may arise where
the E2 learner communicates with either native speakers of the language or
with other F2 users, but clearly the former is more likely in the first type of set¬
ting and the latter in the other two types.

Second language learning in majority language contexts

Second language learners in majority language contexts are typically mem¬

bers of ethnic minorities: immigrants (as in the case of Vietnamese immig¬
rants to the United States), migrant workers (as in the case of Turkish workers
in Germany or Mexican migrants in the United States), or the children of such
groups. These learners vary enormously in the extent to which they approx¬
imate to the language norms of the majority language. In some cases a stable
‘immigrant interlanguage’ (Richards 1972) develops of the kind documented
by Fishman, Cooper, and Conrad (1968) for the Puerto Rican community in
New Jersey, as in this example where the speaker is talking about shopping:

No make any difference, but I like when I go because I don’t have too many
time for buy and the little time we buy have to go some place and I find
everything there.

Such varieties are the product of the social conditions in which the learners
live. The Heidelberger Forschungsprojekt ‘Pidgin-Deutsch’ (1978) refers to
the ‘miserable social situation’ of the foreign migrant worker population in
West Germany which is due ‘not only to economic factors, such as insecurity
of employment, low-prestige work, and so on, but also in large measure to a
rather thorough exclusion from the local social and political life’ (page 2).
One of the main findings of this project is that the length of residence in Ger¬
many functioned as a major explanatory factor in the workers’ acquisition of
F2 German for only the first two years of their stay. After that, it was overrid¬
den by other social factors such as contact with Germans during leisure time,
age at time of immigration, contact with Germans at work, professional
training in the country of origin, and number of years of formal education.
Some groups of learners develop varieties much closer to the target lan¬
guage (for example, Norwegian and Swedish groups in the United States).
How can this be explained? Taylor (1980) identifies three stages in the social
mobility of immigrant groups. Initially there are rewards for maintaining the
218 External factors

LI, as individuals compete for position from within the minority group.
Next, rapid learning of the L2 takes place as individuals identify with the ma¬
jority group and seek to improve their social status. This may lead to what
Lambert (1974) called subtractive bilingualism (see the discussion on page
208 above). Finally, conscious attempts to maintain the LI (the minority lan¬
guage) may be made as individuals react to discrimination by members of the
majority group, whom they perceive as responsible for their lack of social ad¬
vancement. Learners who reach this final stage are likely to achieve additive
bilingualism. How far different groups of learners progress will depend on
the kinds of social factors that the Heidelberger Forschungsprojekt ‘Pidgin-
Deutsch’ identified. These factors govern the degree of contact that takes
place between learners and target language speakers, how useful it is for an
individual learner to make the effort needed to learn the L2, and, in some
cases, the extent to which the interlanguage variety becomes a symbol of eth¬
nic pride.
A somewhat different majority setting arises in countries like Canada and
Belgium where large numbers of the indigenous population learn the lan¬
guage of the majority. In Canada, native French speakers learn English as an
L2, while in Belgium, Flemish speakers learn L2 French (and vice versa).
These settings are of considerable interest because they enable us to compare
the differential levels of proficiency achieved by minority community mem¬
bers learning the language of the majority and, vice-versa, majority members
learning the language of the minority. In both Canada and Belgium, minority
learners of the majority language tend to reach higher levels of proficiency
than majority learners of the minority language. Edwards (1977) found that
French-speaking learners of English in Ottawa maintained their English lan¬
guage skills, whereas English-speaking learners of French tended to lose their
French. Edwards suggests that long-term retention of linguistic and commun¬
icative competence is a function of successful prior learning, opportunity to
use the skills acquired and interest in using them. The French-speaking learn¬
ers reported both more opportunities for using their L2 and greater interest in
doing so than the English speaking ones. Lambert (1974) has claimed that
subtractive bilingualism characterizes many French Canadians learning L2
English, whereas additive bilingualism is more characteristic of English
Canadians learning L2 French.

L2 learning in official language contexts

Under colonial rule the languages of Europe (predominantly English, French,
Portuguese, Spanish, and Dutch) were introduced to a substantial number of
African, American, and Asian countries. The use of these European languages
served to centralize power in the hands of the colonial rulers, who constituted
a social and linguistic elite.1 Independence led some countries to search for
an alternative official language (for example, Hindi in India and Bahasa
 Social factors and second language acquisition 219

Indonesia in Indonesia), but many countries adopted the language of the ex¬
colonial power and maintained it in most of its previous social and official
functions (for example, Nigeria chose English and Zaire chose French, while
India also felt the need to maintain English alongside Hindi). Whether an in¬
digenous or foreign language was chosen as the official language, it consti¬
tuted an L2 for the vast majority of the population. Social and economic
advancement depended to a large extent on its successful mastery—a fact
that constituted a powerful motivation for acquiring it. Official language set¬
tings are often characterized at the national level by the rapid spread of the L2
and at the individual level by rapid acquisition in at least some sections of the
population. In many instances, however, the level of proficiency attained is
limited, as learners fail to develop adequate cognitive academic language pro¬
ficiency (see, for example, Africa’s (1980) account of L2 English proficiency
in the population of Zambia).
Learner communities in official language settings differ from those found
in majority language settings where, as we have seen, some social groups may
resist learning the L2 and seek to maintain their native language as an expres¬
sion of their ethnic vitality. In official language settings there is often less indi¬
vidual resistance to the acquisition of the ‘foreign’ official language, although
there may be considerable group resistance, as in the anti-English movement
in India. Less resistance occurs when the L2 is perceived as an additional lan¬
guage (Fishman, Cooper, and Conrad 1977) rather than a replacement lan¬
guage. In the multilingual situations that are characteristic of newly
independent countries, the foreignness of the L2 is often perceived as desir¬
able, in that it is not associated with any one indigenous group. The choice of
a colonial language as the official language after independence was often po¬
litically motivated by the desire to unite a linguistically and tribally hetero¬
geneous population. For individual learners in countries such as Zambia or
Nigeria, for example, the foreign L2 constitutes a ‘neutral’ language that
could be learnt without any obvious threat to tribal identities.
The choice of national language is, however, a controversial issue. As we
will see when we discuss the different educational contexts of L2 learning, the
choice of an L2 as the medium of instruction in countries such as Zambia can
have a detrimental effect on children’s cognitive and social development.
Also, the use of a foreign language effectively excludes those sections of the
population who do not have access to it (for example, those living in rural
areas with no or few educational facilities) from socio-economic advance¬
ment and, in the eyes of some, it does threaten cultural values. These argu¬
ments have led a number of researchers and educationalists (for example,
Mateene and Kalema 1980; Mateene, Kalema, and Chomba 1985) to pro¬
pose that a vernacular language be used as the official language, either exclus¬
ively or alongside a non-indigenous language.
One the characteristics of L2 learning in official language contexts is the
emergence of new varieties of the target language. Simple varieties analogous
220 External factors

to pidgins (i.e. languages that developed in certain trade and plantation situ¬
ations) appear in certain circumstances, as when the L2 functions as a lingua
franca serving only a limited set of communicative purposes for a group of
speakers. These varieties are similar in form to the ‘immigrant varieties’ dis¬
cussed above. Other varieties can also emerge. Widdowson (1977) suggests
that the opposite of a pidginized variety is ‘babu’, an over-elaborate form of
language that emphasizes how something is said (i.e. expression) over what is
said (i.e. content). He informs us that the term ‘babu’ derives from the name
of the junior Indian clerk whose use of English was ‘characterized by self-
conscious elaboration of phrase which, before the days of the typewriter, was
often conveyed in written form by means of copper-plate handwriting’.
Official language contexts can also give rise to new local standard varieties
for example, ‘New Englishes’ (Kachru 1989). Such varieties are likely to re¬
flect structural features of the speakers’ mother tongue(s) and also overgener¬
alization of rules. However, as Sridhar and Sridhar (1986) and Lowenberg
(1986) argue, it is doubtful whether ‘nativization’ and second language ac¬
quisition involve identical processes. There are differences both with regard
to the linguistic contexts in which apparently similar strategies are employed,
and also in the motivations underlying their use. For example, Lowenberg
(1986: 6) claims that the use of prefixes to coin new lexical items as in ‘outsta-
tion’ (meaning ‘out of town’ in Singaporean English) resembles the kind of
productive process found in established varieties of English rather than a
transitional L2 rule.2 Also, whereas L2 learners view their interlanguage
grammars as transitional and imperfect, users of the new Englishes treat their
grammars as fully developed and display positive attitudes towards them. For
this reason alone, the new standard varieties should not be considered ‘in¬
terlanguages’ in the sense in which this term is generally used.

Second language learning in international contexts

A number of languages—in particular English—are now widely used as in¬
ternational languages. That is, they serve as a means of communication be¬
tween speakers of different languages in a wide range of contexts: business
and trade, academic and scientific, media and the arts, travel and tourism,
and literature. The speakers may or may not be native speakers of the lan¬
guage and the speech events may or may not take place inside a country where
the language is spoken as a mother tongue. Crystal (1988) claims that there
are over 700 million users of English, of whom the majority (400 million) are
non-native speakers. Many of these will use English primarily for interna¬
tional communication. Not surprisingly, the use of English as an interna¬
tional language has attracted considerable attention in recent years (see
Strevens 1980; Quirk 1982 and 1985; Kachru 1982; Smith 1983).
Much of the work on international settings has focused on either describ¬
ing the varieties of language associated with particular contexts of use (for
 Social factors and second language acquisition 221

example, the use of English for air traffic control) or on arguing the merits of
some form of ‘basic’ language which will facilitate communication, teaching
and learning (for example, Quirk’s (1982) Nuclear English and Wong’s
(1934) Utilitarian English). There has been little work on how target-
language varieties associated with international use are mastered by L2 learn¬
ers, or to what extent international use promotes or restricts interlanguage
development. Davies (1989) makes the point that although both interna¬
tional varieties and interlanguages can be described as ‘simplified’, they are in
fact different, as the former involve functional simplification and the latter
formal. Thus, whereas an interlanguage manifests formal reduction—as
when functors are omitted or overgeneralized—and then gradually gets more
complex, international varieties employ standard language forms and are
‘simplified’ only in the sense that they are used to perform a restricted set of
functions. We can speculate that where international varieties are of the very
restricted kind (for example, ‘Seaspeak’ or ‘Airspeak’), they can be mastered
by learning to understand and use a small set of formulas and a limited lex¬
icon, which, as we saw in Chapter 3, are well within the compass of a begin¬
ner learner. In a sense, then, these varieties are both sociolinguistically and
psycholinguistically ‘simple’. We can also speculate that learners who define
their learning task as the mastery of these restricted varieties will achieve only
a limited proficiency. In contrast, learners who engage in international
communication that involves the full resources of the standard language (for
example, the preparation of academic papers for publication) are more likely
to develop a higher level of proficiency.
The use of an L2 in an international setting is characterized by both
non-native speaker-non-native speaker interaction and non-native speaker-
native speaker interaction, but, as Kachru (1986:16) points out, the former is
more common. This has implications for acquisition (see Chapter 7). For
example, Kachru argues that in non-native speaker-non-native speaker
interactions:

the British English or American English conventions of language use are

not only not relevant, but may even be considered inappropriate by inter¬
locutors. The culture bound localized strategies of, for example, politeness,
persuasion and phatic communion ‘transcreated’ in English are more ef¬
fective and culturally significant.

In such situations, then, the learners’ reference group is not speakers of stand¬
ard British or American English but, instead, a prestige local group. Also, in
these situations, we can expect to see creative use made of interlanguage re¬
sources for impression management (see Chapter 5).
222 External factors

Educational contexts

Skuttnab-Kangas (1986; 1988) distinguishes four broad types of educational

contexts: (1) segregation, (2) mother tongue maintenance (language shelter),
(3) submersion, and (4) immersion. These types are all found in multilingual
situations of one kind or another. Here we will also consider a fifth type, for¬
eign language classrooms, as an additional context found in monolingual
situations. The focus of our discussion will be the likely implications of each
context for successful L2 development.
Skuttnab-Kangas identifies a number of factors that she believes will con¬
tribute to educational success in these different settings, grouping them under
four broad headings: (1) organizational, (2) learner-related affective factors,
(3) Ll-related factors, and (4) T2-related factors. Her concern is not specific¬
ally with T2 development but with overall education provision. The factors
that she identifies, however, are all of potential significance for L2 learning.

Segregation
Segregation occurs where the L2 learner is educated separately from the ma¬
jority or a politically powerful minority, who speak the target language as
their mother tongue. Immigrants or migrant workers who are educated in
special schools, centres, or units designed to cater for their language needs
constitute an example of segregation in a majority setting. ‘Bantu education’
in Namibia prior to independence is an example of segregation in a setting
where a powerful minority spoke the official language (Afrikaans) as a
mother tongue.
Skuttnab-Kangas (1988) claims that segregation settings produce poor res¬
ults. She argues that the overall aim of education in these settings is the devel¬
opment of a limited L2 proficiency—sufficient to meet the needs of the
majority or powerful minority and to ensure their continued political and
economic control. She sees segregation education as characterized by inad¬
equate organization (for example, no attempt is made to provide alternative
programmes, and the cultural content of teaching materials may be inappro¬
priate for the pupils) and negative learner-affective factors (for example, a
high level of anxiety and low self-confidence). Although some support for LI
development is provided, this is also usually limited. Negative L2-related fac¬
tors identified by Skuttnab-Kangas include the poor quality of L2 instruction
and the lack of opportunity to practise the L2 in peer-group contexts.
However, the case against segregation is not as clear-cut as Skuttnab-
Kangas makes out. In certain situations, the provision of separate educational
facilities may have beneficial effects. For example, short-term programmes
for refugee populations newly arrived in the United States or European coun¬
tries can help them adjust socially, affectively, and linguistically to the de¬
mands of their new country. It can also be argued that the maintenance of
 Social factors and second language acquisition 223

minority languages requires at least some segregation. Magnet (1990),

for example, draws on the Canadian experience to argue that a minority lan¬
guage will only be viable if its speakers enjoy a ‘degree of autonomy and se¬
gregation in order to develop in their own way’ (1990: 295). The advantages
of segregation are also recognized by minority communities themselves, as il¬
lustrated by their attempts to set up separate schools for their children.
Segregation also has some advantages where L2 learning is concerned. In
particular, because the learners are likely to be at the same level of develop¬
ment, it is possible to tailor input to their level. Where the learners have differ¬
ent Lis, the L2 is likely to serve as a language of classroom communication
and not just as a learning target. This is likely to broaden the functions that it
typically serves. For these reasons, segregation may facilitate the develop¬
ment of ‘survival skills’ in the L2.
In judging segregation it is obviously necessary to consider particular con¬
texts. Skuttnab-Kangas is clearly right to reject segregation when its purpose
is to ensure that learners’ L2 development is restricted, as was the case in
Namibia. However, segregated language education that is designed to meet
the needs of a minority language group and is requested by them may help
them both develop basic L2 skills quickly and also maintain their own LI.

Mother tongue maintenance

Skuttnab-Kangas points out that mother tongue maintenance can take two
forms. In the weaker form, pupils are given classes in their mother tongue,
directed at developing formal language skills, including full literacy. In the
stronger form, pupils are educated through the medium of their mother
tongue. Examples of the former are the programmes for Punjabi established
in Bradford, UK (Fitzpatrick 1987) and for Italian in Bedford (Tosi 1984) for
ethnic minority children living in those cities. Examples of the latter are the
programmes for the seven main language groups in Uzbekistan, and the Finn¬
ish-medium classes for Finnish migrant workers in Sweden (Skuttnab-Kangas
1988). Mother tongue maintenance programmes are based on enrichment
theory, according to which high levels of bilingualism are seen as a cognitive
and social advantage. This contrasts with deficit theory, which views bilin¬
gualism as a burden and as likely to result in cognitive disadvantage. The res¬
ults of research strongly suggest that additive bilingualism (the goal of mother
tongue maintenance) confers linguistic, perceptual, and intellectual advant¬
ages (see Swain and Cummins 1979 for a review).
There is also evidence that mother tongue maintenance settings, particu¬
larly those of the strong kind, result in considerable educational success
(Skuttnab-Kangas 1988). They are characterized by positive organizational
factors (for example, appropriate cultural content in teaching materials), pos¬
itive affective factors (for example, low anxiety, high internal motivation,
224 External factors

and self-confidence in the learners), success in developing full control of the

LI, and a high level of proficiency in the L2.
Mother tongue maintenance provides support for L2 learning in two main
ways. First, ensuring that the L2 is an additional rather than a replacement
language results in learners developing a positive self-identity. As Spolsky
notes, learning an L2 is intimately tied up with one’s personality and being
forced to learn an L2 as a replacement for the LI is a ‘direct assault on iden¬
tity’ (1986: 188). Mother tongue maintenance, then, is more likely to result in
the positive attitudes needed for successful L2 development.
The second way involves a consideration of Cummins’ interdependency
principle (Cummins 1981). This claims that whereas BICS develops separ¬
ately in the LI and L2, CALP is Common across languages.3 Cummins notes
that whereas L2 communicative skills are typically mastered by immigrant
learners in about two years, it can take from five to seven years for the same
learners to approach grade norms for L2 academic skills. The interdepen¬
dency principle has been demonstrated in a number of studies (for example,
Cummins et al. 1984; Verhoeven 1991). Studies of the Portuguese-Canadian
community in Toronto (Cummins et al. 1990), of Japanese immigrant chil¬
dren in Canada (Cummins and Nakajima 1987), and of Turkish immigrant
children in Holland (Verhoeven 1991) support the importance of LI aca¬
demic skills as a basis for successful development of L2 CALP. The notion of
interdependency is an important one because it suggests that the development
of full LI proficiency confers not only cognitive and social advantages attend¬
ant on mother tongue use but also benefits the acquisition of L2 proficiency.

Submersion
Skuttnab-Kangas defines a submersion programme as:

a programme where linguistic minority children with a low-status mother

tongue are forced to accept instruction through the medium of a foreign
majority language with high status, in classes where some children are nat¬
ive speakers of the language of the instruction, where the teacher does not
understand the mother tongue of the minority children, and where the ma¬
jority language constitutes a threat to their mother tongue—a subtractive
language learning situation (1988: 40).

Submersion is common in Britain and the United States, where ethnic minor¬
ity children are educated in mainstream classrooms.
These characteristics of submersion settings are discussed by Cohen and
Swain (1979). Right from the beginning L2 learners are taught with native
speakers. This can create communication problems and insecurity in the
learners. If LI support is provided, it is of the ‘pull-out’ kind, which stigmat¬
izes the L2 child and also deprives learners of the opportunity to progress in
content subjects. Both the content and language teachers are typically
 Social factors and second language acquisition 225

monolingual and thus unable to communicate with the learners in their LI. In
some cases, the learners are actively discouraged from speaking in their LI.
The students’ low academic performance may reflect the low/expectations
that teachers often have of the students, particularly those from certain ethnic
groups (for example, Mexican American students in the United States). Read¬
ing material and subject-matter instruction in the LI are not available, res¬
ulting in increased insecurity in the learners. Parental involvement in the
school programme is usually limited.
For many learners, the disjunction between LI use in the home and L2 use
at school constitutes a painful experience, as Rodriguez’ (1982) autobio¬
graphy illustrates. Rodriguez was the son of a Mexican immigrant who set¬
tled in a mainly white locality of California. At school he was required to use
English exclusively. At home Spanish was spoken, until his parents accepted
the advice of the Catholic nun teachers at his school to speak English. Gradu¬
ally, Rodriguez lost the ability to communicate in Spanish, signalling his re¬
jection of his Spanish-Mexican identity. Although Rodriguez was ultimately
successful in developing a high level of L2 proficiency, this was achieved at
considerable personal and social cost. Rodriguez himself, however, while ac¬
knowledging the discomfort he experienced at both school and home, did not
question the subtractive model of bilingualism to which he was exposed.
Although submersion contexts do not invariably result in lack of success in
learning an L2 (as the Rodriguez’ example demonstrates), in general they do
not facilitate it. Cummins (1988) identifies three characteristics that are im¬
portant for L2 acquisition; (1) a bilingual teacher who can understand stu¬
dents when they speak in their LI, (2) input that has been modified to make it
comprehensible (see Chapter 7 for a discussion of comprehensible input), and
(3) effective promotion of LI literacy skills. Submersion contexts have none
of these. As Cummins notes:

L2 submersion programs for minority students involve virtually no conces¬

sions to the child’s language or culture and have well-documented negative
effects for many children (1988: 161).

Immersion
The term ‘immersion’ has come to refer to a number of different contexts,
which need to be clearly distinguished. Initially, the term was used in the con¬
text of Canadian French immersion programmes, where members of a major¬
ity group (native speakers of English) were educated through the medium of
French, the language of a minority group. There are a number of variants of
these programmes, depending on whether the programme begins early (for
example, in kindergarten) or late (for example, in Grades 4 or 7), and whether
it is full (more or less all instruction is conducted in the L2) or partial (only
part of the curriculum is taught through the L2).
226 External factors

As Cummins (1988) points out, the term ‘immersion’ has also come to be
used to refer to a variety of programmes for minority students. He distin¬
guishes ‘L2 monolingual immersion programs for minority students’ which
provide English-only instruction directed at classes consisting entirely of L2
learners; ‘LI bilingual immersion programs for minority students’, which be¬
gin with LI-medium instruction, introducing L2-medium instruction some
time later; ‘L2 bilingual immersion programs for minority students’, which
emphasize instruction in and on the L2 but which also promote LI skills. He
also notes that, misleadingly, even submersion programmes have been re¬
ferred to as ‘immersion’.
The Canadian Lrench immersion programmes have met with considerable
success. Genesee (1984; 1987) and Swain and Lapkin (1982) review the vari¬
ous programmes, reaching similar conclusions. Immersion students acquire
normal English language proficiency and show the same or better level of gen¬
eral academic development. Lurthermore, immersion students tend to have
less rigid ethnolinguistic stereotypes of the target-language community, and
place greater value on the importance of inter-ethnic contact. These advant¬
ages are evident in ‘disadvantaged’ as well as ‘advantaged’ children. Evalu¬
ation of the different kinds of programmes shows that in general, total
immersion produces better results than partial immersion, and also that early
immersion does better than late.
The Canadian Lrench immersion settings also lead to a high level of L2
Lrench proficiency, particularly with regard to discourse and strategic com¬
petence, where learners achieve near-native-speaker levels (see Swain 1985).
However, such levels are not usually reached in grammatical proficiency and,
as Hammerley (1987; 1989) has pointed out, in some cases a kind of ‘class¬
room pidgin’ can develop.4 Nevertheless, there is now general agreement that
immersion programmes are very effective in promoting L2 development in an
educational setting.
There are many reasons for the success of these majority immersion pro¬
grammes. One undoubtedly has to do with the fact that immersion settings
ensure a plentiful supply of input that has been tailored to the learners’ level
and is therefore comprehensible. There are also social reasons. The learners’
LI and their ethnic identity is not threatened, so it is easy for the learners to
adjust to the immersion setting (see Swain and Lapkin 1985: chapter 6).
Lurthermore, the immersion programmes are optional and, therefore, are
supported by those parents who elect to send their children to them.
Bilingual minority immersion programmes of the kind found in the United
States have been more controversial (see, for example, Epstein 1977 and
Danoff et al. 1978, cited in Cummins 1988). In the United States there has
been considerable opposition to bilingual programmes for linguistic mi¬
norities, as reflected in the Official English Movement (the attempt to have
English designated as the official language of the United States and to ensure
that educational resources are directed towards teaching English rather than
 Social factors and second language acquisition 227

some other language—see Bingaman 1990). Cummins (1988) points outthat

the debate has centred on two arguments, both of which are mistaken. Sup¬
porters of minority bilingual programmes have advanced the ‘linguistic-
mismatch’ argument, according to which minority children will be retarded
academically if they are required to learn exclusively through the L2. This is
mistaken because the French Canadian immersion programmes have shown
conclusively that early instruction through the medium of the L2 has no neg¬
ative effects. Critics of bilingual immersion programs have advanced the
‘maximum exposure’ argument, according to which bilingual education is
detrimental because it deprives learners of the exposure to the L2 necessary
tor successful acquisition. This is refuted by programs which show that mi¬
nority children who spend less time on English while they are developing LI
literacy skills ultimately do just as well in L2 academic skills as those who are
educated exclusively through the L2. Cummins argues that minority pro¬
grammes that are designed in such a way that they reflect the interdependency
principle and the comprehensible input hypothesis have been shown to be
successful. Genesee, however, suggests that the success of minority immer¬
sion programmes also depends on ‘changing the sociocultural fabric of the
school’ (1987: 168-9). He notes that ways are needed to upgrade the status
and power attached to the minority language and to teachers and support
personnel who speak it as an LI. Genesee’s comment points to the need to
consider social as well as organizational factors in immersion education.

The language classroom

Whereas the previous educational contexts have all involved L2 learners in
multilingual settings, the final context we will consider involves ‘the language
classroom’, defined here as a setting where the target language is taught as a
subject only and is not commonly used as a medium of communication out¬
side the classroom. In this sense it includes both ‘foreign’ language classrooms
(for example, Japanese classes in the United States or English classes in China)
and ‘second’ language classrooms where the learners have no contact with the
target language outside the language classroom (for example, ‘ESL’ classes in
a francophone area of Canada).
Two contextual aspects are of potential importance in language classroom
settings according to Gardner and Clement (1990). One concerns the learn¬
ing situation to be found in the classroom. The other is the level of support
which parents give to the foreign/second language programme.
With regard to the classroom learning situation, the role relationships be¬
tween teacher and student are likely to be crucial. In the case of traditional ap¬
proaches to language teaching, where the target language is perceived
primarily as an ‘object’ to be mastered by learning about its formal properties
(see the discussion of ‘formal learning’ on page 214), the teacher typically acts
as a ‘knower/informer’ and the learner as an ‘information seeker’ (Corder
228 External factors

1977b). In the case of innovative approaches where the emphasis is on the use
of the target language in ‘social behaviour’ a number of different role rela¬
tionships are possible, depending on whether the participants are ‘playing at
talk’, as in role play activities, or have a real-life purpose for communicating,
as in information gap activities; the teacher can be ‘producer’ or ‘referee’ and
the learner ‘actor’ or ‘player’. Corder notes, however, that in real-life situ¬
ations outside the classroom, a somewhat different role relationship arises
(‘mentor’ and ‘apprentice’). Thus, even ‘informal learning’ inside the class¬
room may differ from that found in natural settings.5
The nature of these classroom roles is likely to influence the level and type
of proficiency that develops. As we noted earlier, classroom learners often fail
to develop much functional language ability. In part this reflects the predom¬
inance of the knower/information seeker role set in classrooms, although
other factors also play a part (for example, overall contact with the target lan¬
guage). Tessons in French as a second language (FSL) in Canada have been
shown to result in lower levels of L2 proficiency than immersion settings (see,
for example, Lambert and Tucker 1972), although such comparisons are al¬
ways problematic given the difficulty of controlling for intervening variables.
D’Anglejan, Painchaud, and Renaud (1986) also provide evidence of the lim¬
ited success of FSL lessons taught to groups of immigrants, commenting that
‘it is both disappointing yet challenging to discover that after 900 hours of
formal instruction, the vast majority of the subjects have attained proficiency
levels which at best can be described as minimal’ (page 199).
Parents may play an active role by monitoring their children’s curricular
activities. They may also play a more indirect role by modelling attitudes con¬
ducive to successful language learning. A number of studies have found a pos¬
itive relationship between parental encouragement and achievement in L2
learning (for example, Burstall 1975; Gardner and Smythe 1975). Gardner
(1985) argues that parents’ influence on proficiency is mediated through the
students’ motivation.
L2 learning in classroom settings is discussed in detail in Chapters 13
and 14.

Summary

In this section, we have considered the relationship between social context

and L2 learning. The aim has been to identify what are the potential learning
outcomes associated with different types of social contexts, defined in very
broad terms. A basic distinction between ‘natural’ and ‘educational’ settings
has been proposed. There is some evidence to suggest that learners who have
access to natural settings achieve greater functional proficiency than those
who are limited to educational settings. Various types of natural and educa¬
tional settings have been examined. Table 6.2 lists them and provides a sum¬
mary of the potential learning outcomes associated with each. It must be
 Social factors and second language acquisition 229

emphasized once again, however, that the relationship between setting and
learning outcomes is an indeterminate one, as considerable variation is evid¬
ent in each setting, reflecting the interplay of different social factors. Clearly,
there is a need to tease out how these social factors work in much greater
detail than has been possible in this chapter.

Setting Examples Potential learning outcomes

Natural contexts
Majority language settings
monolingual
bilingual
Official language settings
International settings
Educational contexts
Segregation
Mother tongue maintenance
Submersion
Immersion
majority language
minority language
Language classroom
L2 English learnt in USA or UK
L2 English learnt by
Francophones in Canada
L2 English in Nigeria; Bahasa
Indonesian in Indonesia
Use of L2 English for tourism,
business, media etc.
Special migrant worker
programmes in Germany; ‘Bantu
education programmes’1 in
Namibia.
Finnish-medium education for
Finnish minority in Sweden.
Education in mainstream
classrooms for ethnic minority
students in UK and USA;
withdrawal for L2 instruction.
Bilingual education programmes
for English-speaking students in
Canada.
Bilingual education programmes
for Hispanic-speaking students
in the United States.
Foreign language classes in
monolingual countries (e.g.
Japan); Second language ESL
classes for Francophone
students in Canada.
Considerable variation in L2
proficiency:
- immigrant interlanguages
(stable and unstable)
- subtractive bilingualism
- additive bilingualism.
Subtractive bilingualism likely.
L2 learnt as additional language;
Different levels of proficiency:
- pidginized varieties
- ‘babu’
- local standards (e.g. ‘New
Englishes’).
Functionally simplified varieties
(e.g. Airspeak);
Transfer of culture-bound
strategies for impression
management.
L2 proficiency may be restricted
to development of ‘survival
skills’; CALP likely to be
underdeveloped.
High levels of L2 proficiency in
both BICS and CALP.
Low academic performance
resulting from many learners’
failure to develop CALP;
subtractive bilingualism.
High level of functional L2
proficiency but grammatical
proficiency fails to reach NS
levels.
High level of L2 proficiency
achieved if programme attends
to LI literacy and provides plenty
of comprehensible input.
Many learners fail to develop
functional oral L2 proficiency; L2
proficiency higher in reading and
writing skills.

Table 6.2: Social contexts and potential L2 learning outcomes

230 External factors

Social models of L2 acquisition

We will now examine three models of L2 acquisition which seek to account
for the role of social factors. The three models reflect the primary research in¬
terests of their progenitors and the contexts in which they have worked. Two
of the models—Schumann’s Acculturation Model and Giles and Byrne’s
Inter-group Model—have been designed to explain L2 learning in natural
settings, in particular those where members of an ethnic minority are learning
the language of a powerful majority group. The third model—Gardner’s
Socio-educational Model—was derived mainly from studies of L2 learning in
language classrooms, although Gardner argues that it is also applicable to L2
learning in natural settings.

The Acculturation Model

Schumann’s Acculturation Model was established to account for the acquisi¬

tion of an L2 by immigrants in majority language settings. It specifically ex¬
cludes learners who receive formal instruction. Acculturation, which can be
defined generally as ‘the process of becoming adapted to a new culture’
(Brown 1980: 129), is seen by Schumann as governing the extent to which
learners achieve target-language norms. As Schumann puts it:

... second language acquisition is just one aspect of acculturation and the
degree to which a learner acculturates to the target-language group will
control the degree to which he acquires the second language (1978a: 34).

In fact, Schumann (1986) distinguishes two kinds of acculturation, de¬

pending on whether the learner views the second language group as a refer¬
ence group or not. Both types involve social integration and therefore contact
with the second language group, but the first type of learners wish to assimil¬
ate fully into its way of life, whereas the second do not. Schumann argues that
both types of acculturation are equally effective in promoting L2 acquisition.
The model recognizes the developmental nature of L2 acquisition (as dis¬
cussed in Chapter 3) and seeks to explain differences in learners’ rate of devel¬
opment and also in their ultimate level of achievement in terms of the extent
to which they adapt to the target-language culture.
Schumann proposed the Acculturation Model as a means of accounting for
the failure to progress of one of the six learners studied by Cazden, Cancino,
Rosansky, and Schumann (1975). Whereas the other five manifested consid¬
erable development over the ten-month period of the study, Alberto did not
advance in most of the structural areas investigated, for example, negatives,
interrogatives, use of copula ‘be’, and verb auxiliaries (Schumann 1978b).
Alberto’s lack of development could not be satisfactorily explained by either
cognitive development, as he demonstrated normal intelligence, or age, as
many older learners achieve satisfactory levels of L2 proficiency. Schumann
 Social factors and second language acquisition 231

noticed that Alberto’s speech manifested very similar properties to those

found in pidgins, leading Schumann to argue that similar processes of pid-
ginization were at work. Just as pidgins have been noted to develop in contact
situations that call for functionally restricted communication, so learners like
Alberto appear to fossilize because they too have a limited need to communic¬
ate in the L2. According to this view, learners fail to progress beyond the early
stages of acquisition because they require the L2 for only the communicative
function of language (basic information exchange), and not for the integra¬
tive function (social identification) or the expressive function (the realization
of personal attitudes)—see Smith 1972. It was the similarity between
Alberto’s learner language and pidgins that led Schumann to propose accul¬
turation as the primary causative factor.
The extent to which learners acculturate depends on two sets of factors
which determine their levels of social distance and psychological distance
(Schumann 1978a; 1978b; 1978c). Social distance concerns the extent to
which individual learners become members of the target-language group and,
therefore, achieve contact with them. Psychological distance concerns the ex¬
tent to which individual learners are comfortable with the learning task and
constitutes, therefore, a personal rather than a group dimension. The various
social and psychological factors which Schumann identifies as important are
described in Table 6.3. The social factors are primary. The psychological fac¬
tors mainly come into play where social distance is indeterminate (i.e. where
social factors constitute neither a clearly positive nor a negative influence on
acculturation).
A learning situation can be ‘bad’ or ‘good’ (Schumann 1978c). An example
of a ‘good’ learning situation is when (1) the L2 and TL groups view each
other as socially equal, (2) both groups are desirous that the L2 group assimil¬
ate, (3) there is low enclosure, (4) the L2 group lacks cohesion, (5) the group is
small, (6) both groups display positive attitudes towards each other, and (7)
the L2 group envisages staying in the TL area for an extended period. Several
‘bad’ learning situations are possible, as many of the social variables permit
three-way alternatives. Also, different learning situations manifest degrees of
‘badness’ in accordance with the extent of the overall social distance.
In his early writings, Schumann suggested that acculturation affects L2 ac¬
quisition by its effect on the amount of contact learners have with TL speak¬
ers. The greater the contact, the more acquisition takes place. Subsequently,
Schumann (1986) suggests that acculturation may also affect the nature of
the verbal interactions that learners take part in and thus the quality as well as
the quantity of L2 input. The Acculturation Model, however, does not specify
the internal processes that are involved in acquisition. As we will see in Chap¬
ter 9, Andersen (1983b) has added a psycholinguistic dimension to the model
to try to account for these, while Schumann (1990) has also added a cognitive
dimension.
232 External factors

Factor Description

Social distance
1 Social dominance
2 Integration pattern
3 Enclosure
4 Cohesiveness
5 Size
6 Cultural congruence
7 Attitude
8 Intended length of residence The L2 group may intend to stay for a long time or a
Psychological distance
1 Language shock
2 Culture shock
3 Motivation
4 Ego permeability
f boundaries and therefore the extent to which they
The L2 group can be politically, culturally,
technically, or economically superior (dominant),
inferior (subordinate), or equal.
The L2 group may assimilate (i.e. give up its own
lifestyle and values in favour of those of TL group),
seek to preserve its lifestyle and values, or
acculturate (i.e. adopt lifestyle and values of TL
group while maintaining its own for intra-group
use).
The L2 group may share the same social facilities
(low enclosure) or may have different social facilities
(high enclosure).
The L2 group is characterized by intra-group
contacts (cohesive) or inter-group contacts (non-
cohesive).
The L2 group may constitute a numerically large or
small group.
The culture of the L2 group may be similar or
different to that of the TL group.
The L2 group and TL group may hold positive or
negative attitudes towards each other.
short time.
The extent to which L2 learners fear they will look
comic in speaking the L2.
The extent to which L2 learners feel anxious and
disorientated upon entering a new culture.
The extent to which L2 learners are integratively
(most important) or instrumentally motivated to
learn the L2.
The extent to which L2 learners perceive their LI to
have fixed and rigid or permeable and flexible
are inhibited.

Table 6.3: Factors affecting social and psychological distance (based on Schumann
1978b)

The test of any model is whether it is supported by the results of empirical

research. The Acculturation Model has received only limited support, as
Schumann (1986) acknowledges. Maple’s (1982) study of 190 Spanish¬
speaking students enrolled in an ESL programme at the University of Texas
found a strong relationship between social distance and measures of L2 Eng¬
lish proficiency. Seven out of the eight social factors shown in Table 6.3 were
negatively correlated with proficiency. The social factors in descending area
 Social factors and second language acquisition 233

of importance that were found to have a significant effect were: attitudes, so¬
cial class, cohesiveness, intended length of residence, size of L2 group, enclos¬
ure, and perceived status. Maple’s study might be taken as support for the
model except for the fact that the subjects were receiving L2 instruction, and,
therefore strictly speaking, lay outside the model’s frame of reference. Other
studies have failed to support the model either because they found that psy¬
chological distance correlated with advanced proficiency in situations where
social distance was high (for example, Stauble 1978; Kelley 1982), or simply
because no relationship between social distance and development was found
when one might have been expected (for example, Stauble 1984; Schmidt
1983). One of the reasons for these mixed results is the difficulty of measuring
acculturation. Apart from the problem of obtaining reliable measures of each
social factor, there is no principled way of weighting the different variables.
A number of theoretical objections have also been lodged against the
model. Several concern Schumann’s pidginization analogy (see Larsen-
Freeman and Long 1991: 258-9). For example, pidginization is a group phe¬
nomenon while L2 acquisition is an individual one. Pidginization reflects the
incorporation of features from languages other than the target language,
whereas L2 acquisition is always modelled on the target language. However,
these criticisms do not really invalidate the model, as the central construct of
acculturation is not really dependent on the pidginization analogy.
A more serious criticism is that advanced by Spolsky (1989), namely that
Schumann presupposes social factors to have a direct effect on L2 acquisition,
whereas they are more likely to have an indirect one. However, this criticism,
too, is not really justified, as Schumann makes it clear that the effect of the dif¬
ferent social factors is mediated in terms of the amount of contact with target
language speakers that is likely to ensue under different social conditions.
This presupposes that the amount of contact is positively correlated with L2
proficiency. It is not clear, however, to what extent such an assumption is jus¬
tified. The Heidelberger Forschungsprojekt ‘Pidgin-Deutsch’ (1978) reports a
strong relationship between the contact that migrant workers in West Ger¬
many had with native Germans and their syntactic development in the L2.
Similarly, the ZISA Project, also based in West Germany (see Chapter 3)
found clear evidence of a relationship between contact and restrictive simpli¬
fication (the continued use of simplified structures, such as deletion of func¬
tion words, by learners who had developed the ability to use the
corresponding non-simplified structures). In this case both interactive contact
with native speakers at work and in the local neighbourhood, and non¬
interactive contact with the German mass media, were related to more target-
like language use. However, Swain (1981) and Day (1985) both failed to find
a significant relationship between contact with native speakers and L2 profi¬
ciency in studies they carried out in Canada and Hawaii respectively. Freed
(1990), in a study of the effects of contact on the L2 proficiency of 40 under¬
graduate learners of French during a six-week study abroad program, found
234 External factors

that the lower-level students who reported more interactive contact showed
greater gains in grammatical accuracy than those with less. Interactive con¬
tact had no effect or a reverse effect on the grammatical accuracy of ad¬
vanced-level students, however. Non-interactive contact (for example,
listening to the radio, watching television, reading French books and news¬
papers) worked the opposite way; it benefited the advanced but not the lower
level learners. These results testify to the problems inherent in assuming that
‘more contact’ results in ‘more acquisition’. The greatest failing of the Accul¬
turation Model is that it has nothing to say about how social factors influence
the quality of contact that learners experience. In this respect the Inter-group
Model is superior.

The Inter-group Model

We have already seen that Giles and his associates have been primarily con¬
cerned with exploring how inter-group uses of language reflect the social and
psychological attitudes of their speakers. Giles and Byrne (1982), Beebe and
Giles (1984), Ball, Giles, and Hewstone (1984), and Hall and Gudykunst
(1986) have extended this approach to account for L2 acquisition. The inter¬
group theory has become more complex over time in an effort to incorporate
the results of ongoing research about what factors influence inter-group lin¬
guistic behaviour. The account below is based primarily on Giles and Byrne’s
original formulation.
The key construct is that of ethnolinguistic vitality. Giles and Byrne identi¬
fy a number of factors that contribute to a group’s ethnolinguistic vitality (see
Table 6.4). They then discuss the conditions under which subordinate group
members (for example, immigrants or members of an ethnic minority) are
most likely to acquire native-like proficiency in the dominant group’s lan¬
guage. These are: (1) when in-group identification is weak or the LI does not
function as a salient dimension of ethnic group membership, (2) when inter¬
ethnic comparisons are quiescent, (3) when perceived in-group vitality is low,
(4) \yhen perceived in-group boundaries are soft and open, and (5) when the
learners identify strongly with other groups and so develop adequate group
identity and intra-group status. When these conditions prevail, learners ex¬
perience low ethnolinguistic vitality but without insecurity, as they are not
aware of the options open to them regarding their status vis-a-vis native-
speaker groups. These five conditions are associated with a desire to integrate
into the dominant out-group (an integrative orientation), additive bilin¬
gualism, low situational anxiety, and the effective use of informal contexts of
acquisition. The end result is that learners will achieve high levels of social
and communicative proficiency in the L2.
Learners from minority groups will be unlikely to achieve native-speaker
proficiency when their ethnolinguistic vitality is high. This occurs if (1) they
identify strongly with their own in-group, (2) they see their in-group as
 Social factors and second language acquisition 235

inferior to the dominant out-group, (3) their perception of their ethno-

linguistic vitality is high, (4) they perceive in-group boundaries as hard and
closed, and (5) they do not identify with other social groups and so have an
inadequate group status. In such cases, learners are likely to be aware of
‘cognitive alternatives’ and, as a result, emphasize the importance of their
own culture and language and, possibly, engage in competition with the
out-group. They will achieve low levels of communicative proficiency in the
L2 because this would be seen to detract from their ethnic identity, although
they may achieve knowledge of the formal aspects of the L2 through
classroom study.

Variable Description

1 Identification with own ethnic group
2 Inter-ethnic comparison
3 Perception of ethnolinguistic vitality
4 Perception of in-group boundaries
5 Identification with other social groups
This concerns the extent to which learners see
themselves as members of a specific group that
is separate from the out-group, and also
consider their LI an important dimension of their
identity.
This concerns the extent to which learners make
favourable or unfavourable comparisons with
the out-group. Learners may or may not be
aware of ‘cognitive alternatives’.
This concerns the extent to which learners see
their in-group as having low or high status and
as sharing or being excluded from institutional
power.
This concerns the extent to which learners see
their group as culturally and linguistically
separate from the out-group (hard boundaries),
or as culturally and linguistically related (soft
boundaries).
This concerns the extent to which learners
identify with other social groups (occupational,
religious, gender) and, as a consequence,
whether they hold an adequate or inadequate
status within their in-group.

Table 6.4: Variables affecting L2 acquisition according to the Inter-group Model

So far the Inter-group Model is very similar to the Acculturation Model;

both were designed to account for L2 acquisition in majority language set¬
tings, both attempt to specify a set of socio-psychological factors that govern
how successful individual learners will be, and both use these factors to de¬
scribe ‘good’ and ‘bad’ learning situations. However, whereas Schumann’s
model emphasizes ‘contact’ as the variable that mediates between social fac¬
tors and L2 acquisition, Giles and Byrne see ‘interaction’ as crucial. The fac¬
tors they identify determine to what extent learners engage in upward
convergence, and they define L2 learning as ‘long-term convergence’. As we
236 External factors

saw in Chapter 4, much of the work in SLA research based on Giles’ accom¬
modation framework has been directed at discovering which linguistic fea¬
tures are subject to convergence or divergence, and under which interactional
conditions they operate. As such, the Inter-group Model integrates a macro-
and microlinguistic approach to the study of L2 acquisition.
However, the model represents a big leap from Giles’ work on variation to
acquisition, seen as variation over time. The published work has focused only
on the description and explanation of local phenomena in learner language
(for example, Beebe and Zuengler’s (1983) study of Puerto Rican and
Chinese-Thai learners’ use of specific linguistic phenomena in one-to-one in¬
terviews6). As such, there has not really been any real test of the Inter-group
Model. What are needed now are longitudinal studies documenting the social
factors that influence style shifting in L2 learners, the extent to which style
shifting is related to acquisition, and, in particular, whether learners who
regularly converge towards target-language norms in their interactions with
native speakers ultimately achieve high levels of L2 proficiency.

The Socio-educational Model

Gardner’s Socio-educational Model reflects the results of work begun at
McGill University in Montreal in the 1950s and still carried on today. Unlike
the other two models, which were designed to account for the role that social
factors play in natural settings, in particular majority language contexts,
Gardner’s model was developed to explain L2 learning in classroom settings,
in particular the foreign language classroom. It exists in several versions
(Gardner 1979; 1983; and 1985). The following account is derived from the
1985 version.
The model, which is shown schematically in Figure 6.2, seeks to interrelate
four aspects of L2 learning: (1) the social and cultural milieu, (2) individual
learner differences, (3) the setting, and (4) learning outcomes. As such it goes
beyond purely social factors and is more comprehensive than either of the
other two models. The basis of the model is that L2 learning—even in a class¬
room setting—is not just a matter of learning new information but of ‘acquir¬
ing symbolic elements of a different ethnolinguistic community’ (Gardner
1979: 193).
The social and cultural milieu in which learners grow up determines their
beliefs about language and culture. In monolingual settings such as Britain
and the United States, the prevailing beliefs are likely to be that bilingualism is
unnecessary and that assimilation of minority cultures and languages is desir¬
able. In bilingual settings such as Canada both bilingualism and biculturalism
may be encouraged. Gardner identifies a number of variables that result in in¬
dividual difference. The two shown in Figure 6.2 are motivation and lan¬
guage aptitude. As both of these are discussed in detail in Chapter 11, we will
say little about them here. The learners’ social and cultural milieu determine
 Social factors and second language acquisition 237

the extent to which they wish to identify with the target-language culture
(their integrative motivation) and also the extent to which they hold positive
attitudes towards the learning situation (for example, the teacher and the in¬
structional programme). Both contribute to the learners’ motivation,
influencing both its nature (how integrative it is) and its strength. Motivation
is seen as independent of language aptitude (the special ability for learning
languages). Whereas motivation has a major impact on learning in both for¬
mal and informal learning contexts, aptitude is considered to be important
only in the former, although it can play a secondary role in the latter. These
two variables (together with intelligence and situational anxiety) determine
the learning behaviours seen in different learners in the two contexts and,
thereby, learning outcomes. These can be linguistic (L2 proficiency) and
non-linguistic (attitudes, self-concept, cultural values, and beliefs). Learners
who are motivated to integrate develop both a high level of L2 proficiency
andbetter attitudes. The model is dynamic and cyclical.
One of the predictions of the socio-educational model is that the relation¬
ship between the social/cultural milieu and L2 proficiency and also between
learners’ attitudes and their proficiency is an indirect one, whereas that be¬
tween integrative motivation and proficiency is more direct and, therefore,
stronger. Gardner, Lalonde, and Pierson (1983) and Lalonde and Gardner-
(1985) have investigated this using a statistical technique known as Linear
Structural Analysis, which claims to be able to identify causal paths and not
merely correlations among variables. These studies provide support for the
view that factors in the social and cultural milieu are causally related to atti¬
tudes (integrativeness) which in turn are causally related to motivation and
via this to achievement. Other studies have suggested that there may be fur¬
ther intervening variables between setting and L2 achievement. Clement
(1980), for instance, has proposed that there is a ‘secondary motivational
process’ connected with self-confidence. This has led to recent work by
Gardner and his associates on the role of situational anxiety (see Chapter 11).
As we noted earlier, the results of studies that have investigated the rela¬
tionship between integrative motivation and L2 proficiency have been very
mixed. Au (1988) reviews 14 studies carried out by Gardner and his associ¬
ates. Out of these, seven found a nil relationship and four found a negative re¬
lationship between at least some integrative motive measures and L2
achievement. Out of 13 studies conducted by other researchers, only a minor¬
ity produced evidence of even a modest positive relationship. Au argues that
‘there is little evidence that integrative motive is a unitary concept’ (1988:82)
and criticizes Gardner for failing to address why some components are found
to relate to L2 achievement, while others are not. Gardner’s (1988) response
to Au’s criticisms is to claim that it is ‘simplistic’ to assume that the measures
of the different components that make up an integrative motive will be equi¬
valent to each other and refers to other studies not mentioned by Au (re¬
viewed in Gardner 1980 and Lalonde and Gardner 1985), which show that a
238 External factors

Figure 6.2: Gardner’s (1985) Socio-educational Model

composite index of motivation predicts L2 achievement. Gardner also em¬
phasizes the importance of paying careful attention to a number of design and
data analysis factors (for example, the use of valid and reliable measures of
attitudinal and motivational characteristics), with the implication that
studies that have produced nil or negative correlations failed to do this.
The strength of Gardner’s model is that is it explains how setting is related
to proficiency—one of the primary goals of any social theory of L2 acquisi¬
tion—by positing a series of intervening variables (attitudes, motivation, self-
confidence) and by trying to plot how these are interrelated and how they
affect learning. In this respect, it provides the most detailed account of how
social factors influence proficiency currently available. Missing from the
model is any account of how particular settings highlight different factors
that influence attitudes, motivation, and achievement, although Gardner re¬
cognizes the need to pay close attention to the social milieu in order to identi¬
fy alternative factors. Also missing from the model is any reference to the
concept of ‘interlanguage development’ and how this takes place through the
process of social interaction. Gardner’s model, unlike both Schumann’s and
Giles and Byrne’s, only considers ultimate proficiency, measured mainly by
language tests of various kinds. It does not consider the kinds of develop¬
mental patterns documented in Chapter 3. Nor does it consider the social as¬
pects of variability in learner language. The model, therefore, cannot explain
why learners develop in the way they do.

A historical-structural perspective
Finally, we should note that all three models can be criticized from a histor¬
ical-structural viewpoint. This emphasizes the historical background of the
social, political, and economic forces that determine individual choices.
Tollefson (1991) embarks on a lengthy critique of speech accommodation
theory, including Giles and Byrne’s Inter-group Model. His main point is that
it rests on the idea of ‘natural law’ (i.e. that human choice is essentially free
 Social factors and second language acquisition 239

and structures society) and, in so doing, fails to consider the various historical
and structural variables that explain why learners from minority language
backgrounds make the choices they do. According to Tollefson, the concepts
of ethnolinguistic vitality and ethnolinguistic group can only be properly un¬
derstood by considering issues of power and domination in the majority and
minority groups involved. He also argues that by emphasizing the role of con¬
vergence in language learning, the Inter-group Model suggests that ‘learners
who identify with their mother tongue cannot also be fully bilingual’ and
thus, inadvertently, ‘provides a theoretical justification for language educa¬
tion programmes that seek to weaken learners’ ties to their mother tongue
and their community’ (1991: 76). Although Tollefson does not consider the
Acculturation Model and the Socio-educational Model it is clear that they too
are vulnerable to a historical-structural critique.

Conclusion
In this chapter we have explored how social factors affect L2 outcomes. We
have considered the somewhat limited research that has investigated the rela¬
tionship between specific social factors (age, sex, social class, and ethnicity/
attitudes) and L2 proficiency. We have seen that social factors govern the
learners’ choice of reference group, which affects the variety of the target lan¬
guage they choose as their model. We have also examined the different kinds
of social contexts, natural and educational, and how these constrain learning
outcomes. Finally, we have looked at three models of L2 acquisition that em¬
phasize the importance of social factors.
The following are some of the major conclusions to be drawn from this
exploration:
1 Social factors have a general impact on the kind of learning that takes
place, whether informal or formal. Both types can occur in natural and
educational settings but there is a tendency for informal learning to occur
in natural settings and formal in educational settings (particularly in for¬
eign language classrooms).
2 There is no evidence that social factors influence the nature of the processes
responsible for interlanguage development in informal learning. There is
ample evidence to suggest that they affect ultimate L2 proficiency, whether
this is measured in terms of BICS or CALP.
3 The relationship between social factors and L2 achievement is an indirect
rather than a direct one. That is, their effect is mediated by variables of a
psychological nature (in particular attitudes towards to the target lan¬
guage, its culture and its speakers) that determine the amount of contact
with the L2, the nature of the interpersonal interactions learners engage in
and their motivation.
4 Regarding the role of specific social factors, few definite conclusions are
currently possible. There is some evidence to suggest that greater success
240 External factors

in L2 learning will be observed in younger rather than older learners,

females rather than males, middle class rather than working class people
and in learners whose ethnolinguistic vitality is such that learning the L2 is
not perceived as a threat to their ethnic identity. In each case, however, ex¬
ceptions have been reported.
5 Many learners may not be targeted on the standard dialect of the target lan¬
guage. This suggests that measuring progress in terms of learners’ approx¬
imation to the norms of the standard dialect—the normal procedure in
SLA research—may be misleading. Learners may display a degree of
schizophrenia regarding their ‘preference choice’, declaring it to be the
standard variety while actually targeting on some other variety.
6 Some contexts are more likely to result in successful L2 learning than
others (see Table 6.2, page 229).
7 A social model of L2 acquisition will need to consider factors that account
for both the amount and the quality of the contact that learners have with
other speakers of the L2, and for the attitudes that they hold towards the
target language, its culture, and the specific learning situation they find
themselves in.
8 A social theory of L2 acquisition will also need to consider how historical-
structural factors shape individual learner choices.
It is probably true to say that although quite a lot is known about the gen¬
eral impact of social factors on L2 achievement, it is not yet possible to make
accurate predictions. One reason for this is that social factors interact with
other factors of a psychological nature that contribute to individual differ¬
ences (for example, language aptitude, learning style, and personality). The
nature of these interactions is only understood in the broadest terms.

Notes
1 Guy (1988) provides a clear account of how colonial rulers utilized lan¬
guage to establish and maintain their position as social elites.
2 This claim, however, can be challenged. L2 learners make use of a similar
set of communication strategies to native speakers, including word-
coinage (see Chapter 9). A new coinage such as ‘outstation’ might figure in
learner language in the same way it figures in native-speaker speech. Per¬
haps the difference lies not so much in the ‘process’ involved, as Lowenberg
suggests, as in the permanence of the item. In learner language, ‘outstation’
constitutes a temporary solution to a communication problem, with little
chance of it becoming a fixed element in the learner’s lexicon, whereas new
coinages in native-speaker varieties often become fixtures.
3 Contrary to Cummins’ claim, there is also some evidence that BICS is in¬
terdependent. Both Snow (1987) and Verhoeven (1991) show that chil¬
dren’s ability to produce context-embedded language in an L2 matches
their ability to do so in their LI.
 Social factors and second language acquisition 241

4 Hammerley’s attack on the Canadian French immersion programmes has

come in for considerable criticism. Collier (1992: 87), for example,
characterizes his 1989 book as an ‘emotional, polemical, one-sided
account of his personal views ... with scant research evidence cited to
undergird his opinions’.
5 This discussion of roles focuses on the interactional roles adopted by teach¬
ers and learners in the classroom. Such roles reflect the status of the
participants as teachers and students. They reflect the positions which
educational institutions expect them to adopt. These are socially and cul¬
turally determined. This may be why teachers in some African and Asian
countries seem to find it especially difficult to abandon the traditional role
of ‘knower’.
6 In fact, the research reported by Beebe and Zuengler (1983) was completed
before Giles and Byrne (1982) proposed the Inter-group Model. It consti¬
tuted an attempt to reinterpret earlier work by Beebe (see Beebe 1974) in
terms of speech accommodation theory.

Further Reading

A good start is to read C. Baker, Key Issues in Bilingualism and Bilingual

Education (Multilingual Matters, 1988). This has a particularly good section
on attitudes.
Given that much of the literature draws on Cummins’ model of L2 profi¬
ciency, it may also help to read:
J. Cummins, ‘Language proficiency and academic achievement’ in J. Oiler
(ed.), Issues in Language Testing Research (Newbury House, 1983).
A good general article providing a social perspective on L2 acquisition is:
J. Richards, ‘Social factors, interlanguage and language learning.’ Language
Learning (1972) 22.
The following are representative articles dealing with the effect of various
social factors on L2 acquisition:
J. Boyle, ‘Sex differences in listening vocabulary’ Language Learning (1987)
37: 273-84.
N. Holobrow, F. Genesee, and W. Lambert, ‘The effectiveness of a foreign
language immersion program for children from different ethnic and social
class backgrounds: Report 2’ Applied Psycholinguistics (1991) 12: 179-98.
B. Svanes, ‘Attitudes and “cultural distance” in second language acquisition.’
Applied Linguistics (1988) 9: 357-71.
Two important articles dealing with the learner’s choice of target language
variety are:
L. Beebe, ‘Input: choosing the right stuff’ in S. Gass and C. Madden (eds.), In¬
put in Second Language Acquisition (Newbury House, 1985).
L. Goldstein, ‘Standard English: the only target for nonnative speakers of
English?’ TESOL Quarterly (1987) 21: 417-36.
242 External factors

The following are some of the articles mentioned in this chapter dealing
with the relationship between different types of social contexts and L2
acquisition:
T. Skuttnab-Kangas, ‘Multilingualism and the education of minority chil¬
dren’ in T. Skuttnab-Kangas and J. Cummins (eds.), Minority Education
(Multilingual Matters, 1988).
J. Cummins, ‘Second language acquisition within bilingual education pro¬
grams’ in L. Beebe (ed.), Issues in Second Language Acquisition: Multiple
Perspectives (Newbury House, 1978).
A. d’Anglejan, ‘Language learning in and out of classrooms’ in Richards, J.
(ed.), Understanding Second and Foreign Language Learning: Issues and
Approaches (Newbury House, 1978).
A. Davies, ‘Is international English an interlanguage?’ TESOL Quarterly
(1989)23:447-67.
These are the key books and articles dealing with the three social models of
L2 acquisition discussed in this chapter:
J. Schumann, ‘The acculturation model for second language acquisition’ in R.
Gingras (ed.), Second Language Acquisition and Foreign Language Teaching
(Center for Applied Linguistics, 1978).
H. Giles and J. Byrne, ‘An intergroup approach to second language acquisi¬
tion.’ Journal of Multilingual and Multicultural Development (1982) 3:
17-40.
R. Gardner, Social Psychology and Second Language Learning: The Role of
Attitudes and Motivation (Edward Arnold, 1985).
Finally, for those interested in a historical-structural account of the rela¬
tionship between social factors and L2 acquisition, the following is
recommended:
J. Tollefson, Planning Language, Planning Inequality (Longman, 1991).
7 Input and interaction and second
language acquisition

Introduction

Although all theories of L2 acquisition acknowledge the need for input, they
differ greatly in the importance that is attached to it; the role of input in lan¬
guage acquisition is a controversial question. In my previous review of SLA
research (Ellis 1985a: 127ff), I distinguished three different views about its
role: the behaviourist, the mentalist, and the interactionist.
Behaviourist accounts of L2 acquisition propose a direct relationship be¬
tween input and output. Because they reject the idea of ‘mind’ as an object for
inquiry, they ignore the internal processing that takes place inside the learner.
Input is comprised of stimuli and feedback. With stimuli, the person speaking
to the learner models specific linguistic forms and patterns which the learner
internalizes by imitating them. Feedback takes the form of positive reinforce¬
ment or correction, depending on whether the learner’s output is perceived to
be target-like. Behaviourist models of learning emphasize the possibility of
shaping L2 acquisition by manipulating the input to provide appropriate
stimuli and by ensuring that adequate feedback is always available. Acquisi¬
tion is thus controlled by external factors, and the learner is viewed as a
passive medium.
Mentalist theories emphasize the importance of the learner’s ‘black box’.
Although input is still seen as essential for L2 acquisition, it is seen as only a
‘trigger’ that sets off internal language processing (see Cook 1989). Learners
are equipped with innate knowledge of the possible forms that any single lan¬
guage can take, and use the information supplied by the input to arrive at the
forms that apply in the case of the L2 they are trying to learn. As we will see
later, a common assertion of mentalist theories is that the input is ‘indeterm¬
inate’ i.e. the information that it supplies is, by itself, insufficient to enable
learners to arrive at the rules of the target language.
The third type of theory is the interactionist one. This label has been ap¬
plied to two rather different types of theory. According to cognitive interac¬
tionist theories, acquisition is seen as a product of the complex interaction of
the linguistic environment and the learner’s internal mechanisms, with
neither viewed as primary. Cognitive interactionist models of L2 acquisi¬
tion have been drawn from contemporary cognitive psychology and, not
244 External factors

surprisingly perhaps, vary considerably. Despite the differences, however, a

common assumption can be seen to underlie them, namely that input does
have a determining function in language acquisition, but only within the
constraints imposed by the learner’s internal mechanisms.
The second type of interactionist theory is more social in orientation. The
principle that informs these theories is that verbal interaction is of crucial im¬
portance for language learning as it helps to make the ‘facts’ of the L2 salient
to the learner. We will be concerned primarily with social interactionist theo¬
ries in this chapter, delaying consideration of the cognitive kind until Chapter
9. We should note, however, that many cognitive interactionist theories also
see social interaction as the primary mechanism of mental reorganization.
This chapter begins with a brief discussion of the main methods used to
investigate input and interaction. Subsequently it focuses on two major
questions:
1 What are the characteristics of the input that L2 learners typically receive?
2 How does the input influence L2 acquisition?
Question (1) leads us to consider research that has sought to describe L2
input, while (2) leads us to examine research that has sought to explain how
input affects acquisition. Because much of the work on input in second
language acquisition research has followed in the footsteps of work in first
language acquisition research, attention will be given to the latter whenever
this seems appropriate. As there is a separate chapter dealing with input and
interaction in classroom settings (Chapter 13), this chapter will concentrate
on input and interaction in non-instructional settings, although some refer¬
ence to classroom research will be made when this is of particular relevance to
the issue under discussion.

Methods for investigating input and interaction

Many of the earlier input studies (for example, Hatch and Wagner-Gough
1976; Hatch 1978b and 1978c; Peck 1978) used data that had been collected
to study learner language. Because these data consisted of transcriptions of
the interactions in which the learners took part, it was a relatively easy task to
transfer attention from what the learners said to what was said to them. The
data were submitted to both detailed linguistic analyses (for example, to dis¬
cover the frequency with which specific morphemes occurred in the input or
the complexity of input addressed to learners at different stages of
development) and discourse and conversational analysis (for example, to
describe the ways in which topics were nominated and developed or the
strategies used to deal with communication breakdown). Discourse and
conversational
analysis rapidly became the favoured method of analysis, as it provided
researchers with the tools they needed to investigate the nature of the learning
opportunities made available to learners through interaction.
 Input and interaction and second language acquisition 245

Later, experimental and pseudo-experimental studies were designed to in¬

vestigate the effect of specific variables on input and interaction. Studies car¬
ried out by Long (1980a), Gass and Varonis (1985a), and Pica and Doughty
(1985a), among others, have made use of input/interaction data elicited for
the purpose of the enquiry. Long, for instance, in one of the most frequently
cited input studies, asked sixteen pairs of subjects (consisting of American
native speakers of English, and Japanese non-native speakers) to perform six
different tasks, three of them involving one-way information exchange (giv¬
ing instructions, vicarious narrative, and discussing the supposed purpose of
the research) and three involving two-way information exchange (conversa¬
tion and playing two communication games). The advantage of such studies
is that they enable the researcher to manipulate individual variables deemed
likely to influence the quantity or quality of the input provided or to investig¬
ate the effects of the learners’ stage of development on the input provided.
The disadvantage is that it is difficult to determine to what extent the data col¬
lected are representative of the kind of communication the learners typically
take part in. In many cases, as in Long’s study, the interlocutors had no prior
knowledge of each other - a factor which may influence the nature of the in¬
teractions which take place (see the discussion of addressee effects in Chapter
4).
One major advantage of experimental studies is that they make possible the
collection of baseline data, which provide the researcher with some kind of
normative point of reference. Usually these consist of conversations between
native speakers performing the same tasks. Such data enable the researcher to
identify what is special about the input addressed to the learners. However, as
Long (1980a) points out, many studies have failed to collect baseline data.
Not surprisingly, perhaps, introspective techniques have not been widely
used in input research, as neither learners nor native speakers are likely to be
able to comment accurately or reliably on detailed features of the input. How¬
ever, Ferguson (1975) collected data on foreigner-talk by asking students at
Stanford University to rewrite standard English sentences in the way they
thought they would say them to illiterate non-Europeans with no English.
This method suffers from the same drawbacks we noted with discourse com¬
pletion questionnaires in Chapter 5. We cannot be sure that what people
think they would say is what they would actually say.
Another source of introspective information is diary studies. Brown (1985)
used diaries to investigate the kinds of requests for input expressed by L2
learners. She analysed the comments the diarists made in ‘any reference to
input desired, to amount of input given, to type, complexity or meaning¬
fulness of input’ (page 275). Schmidt and Frota (1986) also made use of a
diary study as a means of investigating what one learner noticed in the input.
This study shows that introspection is an important tool for the input
researcher, as it provides one of the best ways of discovering what it is in the
246 External factors

input that learners attend to. It is probably not ‘raw’ input but ‘heeded’ input
that works for language acquisition (see Chapter 9).

The characteristics of input to language learners

Input studies have focused on two issues. A number of studies have examined
what might be called ‘input text’ by trying to establish what native speakers
actually say or write (as opposed to what reference books claim native
speaker usage consists of). Second, there are studies of ‘input discourse’, the
special kind of ‘register’ that is used when speakers address language learn¬
ers.1 Studies of input discourse have been recently extended to include an
examination of the ways in which interaction shapes the input that learners
receive.

Input text: native-speaker usage

A number of researchers have warned of the dangers of making assumptions
about the nature of the input that is addressed to language learners on the ba¬
sis of descriptions of the abstract system of the target language. Lightbown
and d’Anglejan (1985) argue the case for detailed empirical studies of actual
usage, pointing out that such studies are particularly important before any
claims can be made about universals of language acquisition.
A good example of the kind of problem that can arise in SLA research when
the input is assumed rather than attested is found in interrogative structures.
Research reviewed in Chapter 3 provided evidence that learners’ early inter-
rogatives take the form of declarative word order with rising intonation (for
example, ‘You speak English?’). The question arises as to whether this reflects
a transitional construction typical of interlanguage development, indicative
of the process of the learner’s creative construction of L2 rules (Dulay and
Burt 1977). Studies of native-speaker usage suggest that this may not be the
case. Vander Brook, Schlue, and Campbell (1980) and Williams (1990) point
out that many native-speaker questions in English are also non-inverted, par¬
ticularly when there is a high presupposition of a ‘yes’ answer:

A: I’m studying poetry this year.

B: You’re studying poetry this term?

Vander Brook et al. re-examined some of the data from Cancino et al.’s
(1978) study of Spanish learners’ acquisition of English interrogatives and
concluded that in many cases it was not possible to judge the appropriateness
of the learners’ question forms because it was impossible to gauge the amount
of presupposition they were making. In other words, no conclusion regarding
the use of transitional constructions by the learners was possible. Williams’
study of yes/no questions in Singapore English shows how difficult it is to
judge when learners have achieved a level of target-like production without
 Input and interaction and second language acquisition 247

detailed information on what constitutes target-like use. Her advanced level

subjects displayed an ability to produce ‘fully formed questions’ but often did
not do so—certainly to a lesser extent than native speakers of American
English.2
Lightbown and d’Anglejan (1985) found discrepancies between the de¬
clared norms for three French structures (interrogatives, negatives, and word
order) and the use of these structures in everyday spoken French. They found
that uninverted interrogative forms and questions with ‘est-ce-que’ (which
also adhere to declarative word order) are predominant in native-speaker
input, contrary to the claims of the formal descriptions of linguists and
grammarians and also to native-speaker intuitions. In the case of negation,
‘ne’ is often deleted in rapid informal speech (for example, ‘Elle vient pas
avec vous.’), making French a post-negation rather than double-negation
language. They also note that although French is considered to be a subject-
verb-object (SVO) language typologically, spoken French provides copious
examples of alternative orders (VOS, VS and even OSV). This study further
demonstrates the importance of ensuring that learner language is examined
together with accurate information about the input language.
An additional problem facing the researcher is determining which kind of
input to consider when evaluating the learner’s interlanguage. We have al¬
ready seen in Chapter 6 that learners vary in their choice of reference group.
Valdman (1992) talks of the ‘illusive ideal native speaker’, pointing out that
even in cases where there is a single norm (for example, Metropolitan Stand¬
ard French), ‘the norm allows considerable leeway’ (page 84). Valdman goes
on to argue that ‘an invariant TF norm, based on the planned discourse of
educated and cultivated speakers, is an illusory target for learners’ (page 94).
Valdman is more concerned with the problem this poses in choosing appro¬
priate pedagogical norms, but it is also equally problematic for the researcher
concerned with examining the role of input in interlanguage development.
Even assuming that accurate information regarding target-language norms
is available, the problem is not overcome, as in many cases native speakers do
not adhere to these norms when communicating with non-native speakers.
Frequently they modify their input in a number of ways. We will now exam¬
ine the nature of these modifications in input discourse.

Input discourse: the description of modified input

When caretakers speak to young children who are in the process of acquiring
their FI, they typically adjust their speech in a number of ways. The register
that results has been referred to variously as ‘baby-talk’, ‘motherese’, ‘care¬
taker talk’ (the term used in this chapter), and ‘child-directed language’. Simi¬
larly, when native speakers talk to F2 learners they also modify their speech;
the resulting register is known as foreigner talk. It is also possible to talk
about interlanguage talk (see Krashen 1981: 121), the language that learners
248 External factors

address to each other. We will now examine the formal and interactional
characteristics of these registers.

Caretaker talk
Partly as a response to mentalist claims that the input that children receive
from their caretakers is ‘degenerate’ (see Miller and Chomsky 1963), in the
1970s researchers into LI acquisition set out to examine the nature of care¬
taker talk empirically. Important collections of papers were published by
Snow and Ferguson (1977) and Waterson and Snow (1978). Other studies
have continued to be published since, particularly in the Journal of Child
Language. We will concentrate on the main findings of this research.
A number of studies have shown that caretakers adjust their speech form¬
ally so that the input that children receive is both clearer and linguistically
simpler than the speech they address to other adults. Broen (1972) found that
speech addressed to two year-olds has only half the speed used with other ad¬
ults. Garnica (1975) showed that adults use a higher pitch when talking to
children. Sachs (1977) found that mothers tune the pitch, intonation, and
rhythm of their speech to the perceptive sensitivity of their children. Such
modifications are often linked with additional clues provided by gesture and
gaze.
Comparative studies of the speech adults address to other adults and the
speech they address to children have shown that caretakers also make adjust¬
ments in lexis and syntax (Snow 1976; 1977). They use a higher ratio of con¬
tent words to functors and also restrict the range of vocabulary items
employed (i.e. they manifest a low type-token ratio). Modifications in syntax
are evident in a lower mean length of utterance (MLU), a measure that reflects
both the length and the overall linguistic complexity of utterances. Care¬
takers use fewer subordinate and coordinate constructions, and correspond¬
ingly more simple sentences. They avoid sentence embeddings and they
produce sentences which express a limited range of syntactical and semantic
relations.
|
It has been suggested, however, that these characteristics are not necessar¬
ily found in the talk of all caretakers and, in particular, that they may reflect
the particular child-rearing practices of middle-class, English-speaking par¬
ents. Harkness (1977) and Ochs (1982) respectively provide evidence to sug¬
gest that formal simplifications are not the norm in caretaker talk in Kenya
and Western Samoa. Ochs and Schieffelin (1984) argue that many of the
characteristics of talk observed in white, middle-class American caretakers
are a reflection of a cultural predisposition for experts to assist novices and
that this may not be evident in other cultures. Crago (1992) documents how
Inuit children in Arctic Quebec are traditionally expected to learn language
through observing and listening to others rather than through participating in
conversations shaped by questions from adult caretakers.
 Input and interaction and second language acquisition 249

Irrespective of cultural differences, adults’ speech to children has been

shown to be remarkably well-formed, thus refuting the mentalist claim that
the input children receive is degenerate. For example, Cross (1977) found few
examples of speech which was not fluent or intelligible, or of run-on sentences
(in each case, less than 10 per cent of the mothers’ total utterances). Newport,
Gleitman, and Gleitman (1977) reported that only one out of 1,500 utter¬
ances was disfluent.
Another rather different type of adjustment concerns the kinds of topics
that get talked about. Parents tend to follow the here-and-now principle by
talking about topics which can be understood in terms of objects physically
present and actions that are taking place at the time. They avoid talking about
activities that are displaced in time and space until the child has developed the
necessary concepts to understand them. Parents have also been found to pre¬
fer topics that are familiar to their children. Ferrier (1978) has pointed out
that much of the communication with young children centres on routine ac¬
tivities: eating, having a bath, getting dressed, looking at picture books, play¬
ing games, etc. These activities involve caretaker and child in joint attention
on a common set of objects and actions.
Caretaker speech is also characterized by interactional modifications. Both
child and adult caretakers make plentiful use of attention-getters (for ex¬
ample, ‘Look!’ or ‘Hey!’). Adult caretakers make special efforts to ensure that
what they say is understood by their children by frequently checking compre¬
hension and repeating all or parts of their utterances, as in this example from
Snow (1972):

Pick up the red one. Find the red one. Not the green one.
I want the red one. Can you find the red one?

They are also ready to allow the child to initiate and control the development
of topics. Some caretakers appear to be particularly skilful in the strategies
they use to sustain and extend a conversation which their children have
started. Frequently, however, their attempts to communicate are not success¬
ful, either because the caretaker fails to understand what the child has said or
because the child cannot understand the caretaker. In the case of the former
the caretaker is likely to probe further by means of requests for clarification
(such as ‘Mm?’) or requests for confirmation which often take the form of an
expansion of what the caretaker thinks the child has tried to say. When the
child does not understand the caretaker, the caretaker uses repetitions and
paraphrases to sort out the problem. These features, of course, are not unique
to caretaker talk but they have been shown to be especially frequent in com¬
parison to discourse involving adult addressees. As we will see later, the avail¬
ability of semantically contingent input (i.e. input that is closely linked in
meaning to something that the child has already said) has been found to be an
excellent predictor of the child’s rate of progress.
250 External factors

Of considerable interest to researchers is the extent to which the adjust¬

ments are ‘fine-’ or ‘rough-tuned’. Krashen (1980) defines fine-tuning as the
provision of the specific linguistic features which the child is ready to acquire
next and claims that caretaker-talk is characterized only by rough-tuning. As
evidence he cites the positive but ‘not strikingly high’ correlations between
linguistic input complexity and linguistic competence in children found by
Newport, Gleitman, and Gleitman (1977). Krashen is surely right in claiming
that caretakers do not aim their input at specific linguistic structures, but he
probably underestimates the extent to which adult caretakers are sensitive to
their children’s progress. As we will see when we discuss the relationship be¬
tween input frequency and acquisition, Wells (1985) provides evidence to
suggest that adults step up the frequency of specific linguistic features in their
input shortly before their children first use them in their own speech. Cer¬
tainly, there is plenty of evidence to show that adjustments in caretaker talk
occur on a continuous scale; that is, they are responsive to the development
evident in individual children at different stages of acquisition. Cross (1977)
considers that this justifies claiming that the input is fine-tuned. ’
Researchers have also considered the purposes served by caretaker talk.
Ferguson (1977) suggests three possible functions: (1) to aid communication,
(2) to teach language, and (3) to socialize the child. It is the first of these that
seems to be the most important. Caretakers seek to communicate with their
child and this leads them to modify their speech in order to facilitate the ex¬
change of meanings. As Brown (1977: 26) put it, the primary motivation is to
communicate, to understand and to be understood, and to keep two minds
focused on the same topic. Brown and Hanlon (1970) have shown that moth¬
ers do correct children to ensure that what they say is true.4 Thus, if a child
mislabels an object (for example, refers to a horse as ‘doggie’), the mother is
likely to respond with either an explicit correction (such as ‘No, it’s a horsie’)
or an implicit correction (for example, ‘Yes, the horsie is jumping’). Mothers
also pay attention to their children’s pronunciation of words and draw their
attention to politeness formulae (for example, when to say ‘thank you’).
However, they pay little attention to the grammatical correctness of their
children’s speech, allowing even blatant errors to go uncorrected. In general,
then, if caretaker talk serves to teach syntax and to socialize the child into the
adult’s culture it does so only indirectly, as offshoots of the attempt to
communicate.5
How do caretakers determine the nature and the extent of the modifica¬
tions that need to be made? Gleason and Weintraub (1978) suggest that they
must form a general idea of children’s linguistic ability, particularly their abil¬
ity to understand. They argue that they do not have an accurate knowledge of
the specific linguistic features the children have mastered. Of crucial import¬
ance is the extent to which children comprehend what is said to them and the
extent to which they signal their comprehension or lack of it to their care¬
takers. Children tend to become inattentive when they do not understand,
 Input and interaction and second language acquisition 251

causing their caretakers to modify their speech until attention is restored.

L^ross (1977; 1978) found little evidence that the mothers in her study were
able to monitor their own or their children’s syntactic levels, but she did find
that they evaluated the extent to which the children were able to demonstrate
an understanding of their previous utterances. The fact that caretaker modi¬
fications have been observed in the speech that even four-year-old children
address to their younger siblings (see Shatz and Gelman 1973), also suggests
that they must be based on a very general assessment of the learner’s com¬
municative abilities.
In summary, there is now plentiful evidence to show that the speech that
caretakers address to children is well-adapted to their linguistic abilities, par¬
ticularly comprehension. Three main features of caretaker talk stand out: (1)
it is more grammatical than speech addressed to adults, (2) it is simpler, and
(3) it is more redundant. However, as Snow (1986) has pointed out, it cannot
be concluded on the basis of these findings that there is no innate component
in T1 acquisition, as the studies do not enable us to determine the relative im¬
portance of innate and input factors. The studies show only that the claim
that children receive degenerate input is unwarranted.

Foreigner talk
The foreigner talk (FT)6 used by native speakers when communicating with
non-native speakers displays many of the characteristics of caretaker talk.
There are also some differences, however, particularly when the non-native
speakers are adults. Freed (1980; 1981) compared the speech of fifteen moth¬
ers studied by Newport (1976) with that of native speakers of American Eng¬
lish to eleven adult non-native speakers. She found no differences in the
degree of well-formedness and syntactic complexity (although instances of
ungrammatical FT are not uncommon—see page 253), but she did find some
in the distribution of sentence types and the interpersonal functions they
encoded. In particular, declaratives were much more common in the
foreigner talk, and yes/no questions and imperatives less common. Freed
suggests that this reflects a general difference in purpose: whereas the main
functional intent of caretaker talk is that of directing the child’s behaviour,
that of foreigner talk is the exchange of information. It should be noted, how¬
ever, that when FT is addressed to young children, it appears to resemble
caretaker talk fairly closely. Hatch, Peck, and Wagner-Gough (1979)
analysed the input in Huang’s (1970) study of a five-year-old learner and
found that imperatives and questions far outweighed declaratives. The
crucial factor, therefore, may be age.
A detailed study of foreigner talk necessitates a consideration of a number
of issues: (1) the extent to which FT is grammatical and ungrammatical, (2)
the nature of the formal modifications found in FT, (3) the nature of the inter-
252 External factors

actional modifications found in FT, (4) discourse structure in FT, and (5) the
functions served by FT.

1 Ungrammatical input modifications

In one of the earliest discussions of FT, Ferguson (1971) noted that in lan¬
guages where native speakers employ a copula in equational clauses in nor¬
mal communication (for example, ‘Mary is a doctor’) they often omit it in
talk directed at foreigners. Ferguson suggests that this is because the absence
of the copula is considered simpler than its presence. The omission of copula
is a clear example of ungrammatical FT. In subsequent publications (Fergu¬
son 1975; Ferguson and Debose 1977), Ferguson suggests that ungrammat-
icality is evident in three ways: (1) omission of grammatical functors such as
copula, articles, conjunctions, subject pronouns, and inflectional morpho¬
logy, (2) expansion, as when ‘you’ is inserted before an imperative verb (for
example, ‘You give me money.’), (3) replacement/rearrangement, as when
post-verbal negation is replaced by pre-verbal negation in English FT (for
example, ‘No want play’). Frequently utterances will manifest all three types
of ungrammaticality. As Ferguson (1971) notes, many of the features found
in FT are also evident in pidgins.
A number of studies provide evidence of ungrammatical FT (see Long
1980a and Larsen-Freeman and Long 1991 for reviews). It is particularly
likely in what Lerguson and Debose (1977) refer to as ‘talking down’ situ¬
ations. Thus, Clyne (1978) reports finding examples in the speech that Aus¬
tralian factory foremen use to address foreign workers. Germans have been
found to address guest workers in the same way (Heidelberger Lorschungs-
projekt 1978). However, it is also been found in situations of a more neutral
kind, as when passers-by give directions to tourists (Walter-Goldberg 1982,
cited in Larsen-Freeman and Long 1991) and even in conversations between
friends (Hatch, Shapira, and Wagner-Gough 1978).
There are striking similarities between ungrammatical FT and learner lan¬
guage. Table 7.1 compares the speech of an adult learner (Zoila) with that of
a native-speaker friend (Rina) in the same conversation. It shows that Rina’s
input matched Zoila’s output in a number of ways. Hatch, Shapira, and
Wagner-Gough (1978) observe that Rina felt unable to stop herself produc¬
ing ungrammatical utterances but they also point out that her speech was not
an exact copy of Zoila’s. The similarity between FT and learner language
should not be taken as evidence in favour of the matching hypothesis (i.e. that
the source of learners’ ‘errors’ is ungrammatical FT), for, as both Long
(1983a) and Meisel (1983) have noted, it may reflect a common set of cognit¬
ive processes (see the discussion on page 265 later in this section). However,
native speakers may well introduce ungrammatical forms of the kind they
observe in learner language into their speech as part of the process of
 Input and interaction and second language acquisition 253

Grammatical structure Learner language Foreigner talk

Copula Regularly deleted - but does Only deleted in 5 out of 43

occur in some contexts.
Pronoun ‘it’ Pervasive deletion
Verb tense Uninflected verb form used
for all time reference; also
V -ing used in apparent free
distribution. No aux-do.
Negatives ‘no + V’ and ‘1 don’t know’
used.
Possessives No possessive’s.
‘For’ ‘For’ used in expressions like
‘Is upset for you?’
instances.
Also pervasively deleted.
V -ing used grammatically.
Aux-do regularly deleted.
Mixed ‘no + V’ and
grammatical negatives used;
‘no + V -ing’ negatives most
common.
Nouns consistently marked
with possessive’s.
Similar use of ‘for’ found in 3
out of 15 instances.

Table 7.1: A comparison of learner language and foreigner talk (based on data
from Hatch, Shapira, and Wagner-Gough 1978)

accommodating to their addressee (see Chapters 4 and 6 for a discussion of

accommodation theory).
Non-standard forms are also evident in other levels of language: pronunci¬
ation and lexis, for instance. Epenthesis (the insertion of an additional
vowel), the replacement of reduced vowels by full vowels, and exaggerated
intonation similar to the kind observed in caretaker talk have all been noted
in FT pronunciation. Interestingly, though, native speakers practising FT do
not seem to follow caretakers in using a higher pitch. In lexis, ‘ungrammat¬
ical’ adaptations include the use of names in place of pronouns, a special lex¬
icon of quantifiers, intensifies and modal particles and, in highly marked
forms of FT, the use of foreign or foreign-sounding words such as ‘amigo’,
‘capito’, and ‘compris’, which Meisel (1980) reports observing in German
FT.
A number of factors appear to induce ungrammatical FT. Fong (1983a)
suggests that four factors may be involved:
1 The learner’s level of proficiency in F2—ungrammatical FT is more likely
when the learner’s proficiency is low.
2 The status of the native speaker—ungrammatical FT is more likely when
the native speaker is or thinks he or she is of higher status.
3 The native speaker has prior experience of using FT but only of the ‘limited
kind’ used to address non-native speakers of low proficiency.
4 The extent to which the conversation is spontaneous—ungrammatical FT
is less likely in planned, formal discourse or in experimental situations.
254 External factors

However, ungrammatical FT can occur both with interlocutors who are fa¬
miliars (as in the conversations between Rina and Zoila) and with strangers,
suggesting that factors other than those listed by Long are at work. It is, in
fact, not yet possible to identify the exact conditions that will result in un¬
grammatical FT, perhaps because native speakers vary both culturally and in¬
dividually in the kind of FT they prefer to use.
As Meisel (1980) points out, ungrammatical FT is generally felt to imply a
lack of respect. Meisel reports that Italian and Spanish workers in Germany
reacted negatively when played recordings of speech samples containing un¬
grammatical FT, claiming that it showed contempt on the part of the native
speakers. However, it may not be the presence of ungrammatical modifica¬
tions per se that arouses negative responses in learners, but their awareness of
being addressed in a special manner. Lynch (1988) found that some learners
objected to the speech of a teacher because they perceived it as ‘talking down’,
even though it did not contain obvious ungrammatical modifications. Also, in
cases of very close relationships (as that between Rina and Zoila) ungram¬
matical adjustments do not appear to be objected to.

2 Grammatical input modifications

Ungrammatical FT is highly marked. In many situations it does not occur,

suggesting that it constitutes a particular discourse type. Arthur et al. (1980)
recorded sixty telephone conversations between adult non-native speakers of
English and airline ticket agents and reported no instance of ungrammatical
input modifications. Studies of teacher talk (for example, Henzl 1973 and
1979; Gaies 1977; Hakansson 1986) also, not surprisingly, report an absence
of ungrammatical modifications, although other studies (for example, Hatch,
Shapira, and Wagner-Gough 1978) did find instances in the language that
teachers use to organize and manage classroom activities. Grammatical FT is
the norm in most classrooms, however.
Grammatical FT is characterized by modifications indicative of three gen¬
eral, processes: (1) simplification, (2) regularization, and (3) elaboration. To
illustrate the differences in the FT that results from these three processes, con¬
trived examples are provided in Table 7.2. It should be noted that whereas (1)
involves an attempt on the part of native speakers to simplify the language
forms they use, (2) and (3) are directed at simplifying the learners’ task of pro¬
cessing the input and can, in fact, result in the use of language that is not al¬
ways simple in itself. This is important because it means that FT provides not
only simple input, corresponding perhaps to what learners already know, but
also input containing linguistic features that they have not yet learnt.
The account of FT that follows draws on a number of studies: Gaies 1977;
Arthur et al. 1980; Long 1981a and 1981b; Scarcella and Higa 1981; Hatch
1980 and 1983b; Chaudron 1982; Kleifgen 1985; Hakansson 1986; Wesche
and Ready 1985; Griffiths 1990 and 1991a.
 Input and interaction and second language acquisition 255

Type of speech Example

Baseline (i.e. speech The 747 is a large-sized jet manufactured by Boeing, an

addressed to native American company, with a seating capacity of over 500,
speakers) arranged on two decks.
Simplified FT Well, er,... the 747’s a big jet. And er... er... it’s a Boeing, an
American plane. Er... there’s over 500 seats with er... some on
top and er... some down below.
Regularized FT The 747, it is a big jet. It is made by Boeing which is an
American company. The seats, they are on two levels. There is
a top level and a bottom level.
Elaborated FT The Boeing 747 or jumbo, as it is called, is a very large jet,
manufactured or made by an American company, a firm in
Seattle USA. It has the capacity or space to seat a large number
of passengers, over 500 people. The seats are on two decks or
levels, some up on top and some down below.

Table 7.2: Contrived examples of FT illustrating the processes of simplification,

regularization, and elaboration

Table 7.3 summarizes the main linguistic modifications that have been
claimed to contribute to simplification. One way of simplifying is by ad¬
justing temporal variables such as speech rate (measured usually in syllables
per second), articulation rate (measured by calculating the ratio of the total
number of syllables to the total articulation time) and silent pause phenom¬
ena (pause duration, pause distribution, and pause frequency). Griffiths
(1991a) summarizes the research in this area of FT. His survey shows that al¬
though temporal variables have been frequently commented on in the literat¬
ure (for example, Freed 1981; Dudley-Evans and Johns 1981; Hatch 1983b;
Klein 1986), these comments are not accompanied by reports of detailed em¬
pirical results. Henzl’s (1973; 1979) study of the teacher talk in three different
languages (Czech, English, and German) addressed to advanced and beginner
L2 learners and native speakers, does give results, however. The study shows
that the teachers adjusted their speech rate in accordance with the listeners’
proficiency, as does Hakansson’s (1986) study of Swedish teachers’ class¬
room talk. Griffiths points out a number of methodological flaws in both
studies (for example, failure to control for the sequence of audiences and to
account for classroom activity silences) and also draws attention to the fact
that in Henzl’s study there was enormous variability from one language to
another and, in both studies, from one teacher to another within the same
language. In the case of syntactic and lexical modifications, simplification
is achieved both by avoiding difficult items in the target language and also by
reduced use of them.
Regularization entails the selection of forms that are in some way ‘basic’ or
‘explicit’. Examples include: fewer false starts, the preference for full forms
256 External factors

Type of simplification Comment

1 Temporal variables
2 Length
3 Syntactic complexity
4 Vocabulary
Speech to NNSs is often slower than that addressed to NSs
- mainly as a result of longer pauses.
FT makes use of shorter sentences (fewer words per T-unit).
FT is generally less syntactically and propositionally
complex, i.e. fewer subordinate clauses of all kinds
(adjectival, noun, and adverbial), greater use of parataxis
(e.g. simple coordinate constructions), and less preverb
modification.
FT manifests a low type-token ratio and a preference for
high frequency lexical items.

Table 7.3: Simplification in grammatical foreigner talk

over contracted forms; the preference for canonical word order noted by
Long, Gambhiar, Ghambiar, and Nishimura (1982) in English, Hindi, and
Japanese; the use of explicit markers of grammatical relations (for example,
‘He asked if he could go’ rather than ‘He asked to go’); the movement of top¬
ics to the front of sentences (for example, ‘John, I like him’); the avoidance of
forms associated with a formal style (for example, ‘tu’ is preferred to ‘vous’ in
French FT); and the avoidance of idiomatic expressions and the use of lexical
items with a wide coverage (for example, ‘flower’ rather than ‘rose’). Hatch
(1983b: 66-7) suggests a number of ways learners might benefit from regular¬
izations of these kinds. For example, they help to make the meanings of utter¬
ances more transparent. This may be achieved by increasing the processing
time available to learners or by making key structural elements more salient,
thereby helping them to identify constituent boundaries in utterances.
Elaboration is the opposite of simplification, but to claim that FT evidences
both is not contradictory, as both processes can occur at different times. Elab¬
oration often involves lengthening sentences in an attempt to make the mean¬
ing clear. Native speakers often use analytic paraphrases of lexical items they
consider difficult. Chaudron (1983b) provides a number of examples in the
speech used by a university lecturer to a class of ESL learners (‘hold on tightly’
is used in place of ‘cling’; ‘there’s still this feeling...’ instead of ‘we have this
myth’). Native speakers also sometimes offer synonyms (‘funds or money’)
and they define items. They may add information that helps to contextualize
an item (for example,’If you go for a job in a factory...’ where ‘in a factory’ is
redundant). Chaudron suggests that such elaborations are designed to make
the message more ‘cognitively simple’, but he also makes the point that they
can result in too much redundant and confusing information. The lecturer he
studied sometimes over-elaborated, making the interconnections between
ideas difficult to comprehend.
 Input and interaction and second language acquisition 257

As in caretaker talk, these different kinds of adjustments are continuous in

nature; speakers make more or less of them depending on their perception of
the learner’s ability to understand. Hakansson (1986), for instance, found
that the input provided by a teacher to learners of L2 Swedish increased in
length over time as a result of an increased use of subordinations and an ex¬
pansion of nominal phrases through increased modification. Kleifgen (1985)
found that a kindergarten teacher’s input became more complex over time to
children who showed improvement, but remained static for those who did
not.
The learners’ proficiency is not the only factor that determines the extent of
modification. Age has an effect as well. Scarcella and Higa (1981) suggest that
linguistic simplification is in fact triggered more by age than linguistic com¬
petence, on the grounds that they found much more to occur in FT addressed
to children aged between eight and ten years than in FT to adults.
The linguistic modifications that have been observed to occur in FT corres¬
pond closely to the language found in simplified texts of the kind found in
graded readers. As such, the input is often considered ‘non-authentic’ (see
Vincent 1986). It is worth noting, however, that in FT the modifications oc¬
cur quite spontaneously as part of the process of trying to communicate with
learners of limited competence.

3 Interactional modifications

Increasingly, studies of foreigner talk have switched their attention from lin¬
guistic to interactional modifications. This has been motivated in part by the
finding that interactional modifications occur even when input modifications
do not (Long 1980a) and also by theoretical claims regarding the importance
of this type of modification for comprehension and acquisition (see following
sections in this chapter).
A useful distinction can be made between those interactional modifications
that involve discourse management and those that involve discourse repair.7
The former are motivated by the attempt to simplify the discourse so as to
avoid communication problems, while the latter occur when some form of
communication breakdown has taken place or in response to a learner utter¬
ance that contains an error of some kind (factual, linguistic, or discourse).
These distinctions are shown diagrammatically in Figure 7.1 on page 258.

Discourse management

One of the most effective ways of managing discourse with native speakers is
to ensure that the topic of the conversation is understood. Long (1983a)
identifies a number of strategies which native speakers use to achieve this end:
selecting salient topics, treating topics simply and briefly, making new topics
salient, and, when necessary, relinquishing topic control, although this last
strategy seems to relate more to discourse repair than discourse management.
25 8 External factors

discourse management
- amount and type of information conveyed
- use of questions
- here-and-now orientation
- comprehension checks
interactional
- self-repetition repair of communication
modifications
breakdown

- negotiation of meaning
(requests for clarifica¬
tion; requests for
discourse repair confirmation; self- and
other-repetitions)
- relinquishing topic

repair of learner error

- avoidance of other-
correction
- on-record and off-
record corrections

Figure 7.1: Types of interactional modifications in foreigner talk

One method used by native speakers to control topic concerns the amount
and type of information that is communicated. Arthur et al. (1980) compared
the number of ‘information bits’ that native speaker airline agents included in
their answers to a specific telephone enquiry from native speakers and non¬
native speakers (‘What kind of plane is a_?’). A distinction was made be¬
tween ‘simple’ information (such as ‘size’ and ‘jet’) and ‘complex’ informa¬
tion (for example, ‘seating capacity’, ‘name of manufacturer’, and ‘seating
arrangement’). There was no difference in the amount of simple information
given to native speaker and non-native speaker callers, but significant differ¬
ences were found in the amount of complex information, the non-native
speakers receiving far less. Derwing (1989) found that native speakers ad¬
justed the information they provided about a film they had seen when speak¬
ing to low-proficiency L2 learners. The information contained in the
speakers’ propositions was classified as belonging to one of three categories:
(1) crucial information, (2) non-essential major information, and (3) minor
information, consisting of background or irrelevant information. There was
no difference in the amount of crucial information which the native speaker
and non-native speaker addressees received, but differences were evident in
the relative proportions of major and minor information. Overall the nar¬
rators included less major information and more minor information in speech
to the learners. There were considerable individual differences, however,
which, as we will see later, had a significant effect on the non-native speakers’
comprehension. Ehrlich, Avery, and Yorio (1989) also found evidence of
variation in the amount of information supplied by individual native speakers
 Input and interaction and second language acquisition 259

in a problem-solving task. They distinguish ‘skeletonizing’, where the barest

details are provided, from ‘embroidering’, where the information is expanded
and embellished. Interestingly, they found both strategies present in inter¬
actions with both native speakers and non-native speakers, suggesting that
individual native speakers may have preferred interactional styles which they
use irrespective of their interlocutors. It is clear, though, that often native
speakers do seek to manage discourse with non-native speakers by regulating
the amount and type of information they provide.
Native speakers also make use of questions to establish and control topics.
We have already noted that questions appear to be more frequent in FT than
in caretaker talk. Long (1981b) found that in conversations between native
English speakers and elementary level Japanese learners, the native speakers
initiated most of the topics, typically making use of questions to do so. Ninety
six per cent of all topic initiations were questions, whereas in native speaker-
native speaker conversations only 62 per cent were. Long suggests a number
of reasons why questions were favoured: they compel answers, they signal to
the non-native speaker that a turn is approaching, and they lighten the learn¬
er’s conversation burden because they encode part (and sometimes all) of the
propositional content required to respond. Long also found differences in the
types of questions used. In conversations with the learners, the native speak¬
ers made greater use of yes/no and ‘or’ type questions. One possibility is that
the use of questions in Long’s study reflected, in part at least, the lack of fa¬
miliarity between the participants who made up the pairs. However, Gaies
(1982) obtained similar results, suggesting that this was not the case. In this
study, unlike Long’s study, the native speakers and non-native speakers were
already acquainted with each other, but the native speakers still used more
questions when talking with the non-native speakers. It should be noted,
though, that in all these studies the native speakers were adults; child native
speakers seem less inclined to establish and develop topics through ques¬
tioning (see Hatch 1978b; Peck 1978).
A third strategy of discourse management is to select topics that have a
here-and-now orientation (see the discussion of this feature of caretaker talk
on page 249). Long (1980a) and Gaies (1982) both report significantly more
present-tense verbs in native speaker speech addressed to non-native speakers
than in speech addressed to other native speakers, suggesting such an orienta¬
tion. The here-and-now orientation allows learners to make use of the imme¬
diate context to interpret the meaning of utterances.
Finally, native speakers have been noted to tty to manage discourse by fre¬
quently checking whether the learner has understood. Comprehension
checks (for example, ‘You understand?’, ‘Okay?’) have been found to occur
more frequently in native speaker-non-native speaker discourse than in
native speaker-native speaker discourse (Long 1981a; Scarcella and Higa
1981 i. Teacher talk, in particular, seems to be rich in comprehension checks.
260 External factors

Pica and Long (1986) found that ESL teachers were much more likely than
native speakers to check comprehension in informal conversations.

Discourse repair

The need for discourse repair arises in certain types of problematic commun¬
ication. Gass and Varonis (1991) present a taxonomy of problematic com¬
munication types (see Figure 7.2). An initial distinction is made between
‘non-engagement’ and ‘miscommunication’. The former occurs either when
there is ‘non-communication’ (for example, when a non-native speaker
avoids talking to a native speaker) or when there is ‘communication breakoff’
(for example, when a native speaker stops communicating as soon as they dis¬
cover they are talking to a non-native speaker). ‘Miscommunication’ occurs
when some message other than that intended by the speaker is understood. It
can take the form a ‘misunderstanding’ or an ‘incomplete understanding’
(either ‘non-understanding’ or ‘partial understanding’), depending on
whether or not the participants overtly recognize a problem and undertake
repair. In the case of an ‘incomplete understanding’ remediation occurs, but
in the case of a ‘misunderstanding’ no repair occurs and the speakers are
likely to lapse into silence. Gass and Varonis also note that miscommunica¬
tion can occur both as a result of cross-cultural differences in the way lan¬
guage is interpreted and because of purely linguistic difficulties.

Problematic communication

Non-engagement Miscommunication

Non-commun- Communication Misunderstanding Incomplete

ication breakoff understanding

, non-under- partial
standing under¬
standing

Figure 7.2: Problematic communication types (from Gass and Varonis 1991)

Repair, then, occurs when there is an ‘incomplete understanding’. It takes

the form of negotiation of meaning - the collaborative work which speakers
undertake to achieve mutual understanding. Native speakers typically use re¬
quests for clarification (‘Sorry?’, ‘Huh?’, ‘I beg your pardon’) and requests for
confirmation, which often make use of intonation or tag questions:

NNS: Mexican food have a lot of ulcers.

NS: Mexicans have a lot of ulcers?
(Young and Doughty 1987).
 Input and interaction and second language acquisition 261

Other conversational modifications that help to repair discourse are self-

and other-repetitions, which can be exact or semantic (i.e. paraphrases) and
complete or partial. It should be noted that not all repetitions have a repairing
function; as Pica and Doughty (19 8 8) point out, native speakers also use them
to manage discourse (i.e. to try to prevent a communication problem arising).
Native speakers may also abandon the attempt to negotiate meaning by giv¬
ing up a topic and allowing the non-native speaker to nominate an alternative
one.
A considerable amount of research has been undertaken to establish the
conditions that promote the negotiation of meaning, motivated by the claim
that the comprehensible input that can result from it is of particular benefit to
L2 acquisition (see the section on input and acquisition in this chapter). Pica
(1987) argues that the most important factors concern the social relation¬
ships between the interactants. Interaction involving participants of equal
status ensures that ‘learners and their interlocutors share a need and desire to
understand each other.’ Conversely, unequal status makes it difficult and
even unnecessary for participants to restructure interaction. This claim mir¬
rors that made by Wells and Montgomery (1981) for caretaker interaction.
When mothers act as conversational partners plentiful negotiation takes
place, but when they adopt a tutorial role it is inhibited.
The distribution of power among the participants helps to explain a num¬
ber of other findings, such as the influence that the non-native speaker’s age
has on FT. Scarcella and Higa (1981) found that adult native speakers as¬
sumed much greater responsibility for conversations when speaking with
non-native speaker children than with non-native speaker adults. With chil¬
dren they tended to dominate conversations so that less negotiation took
place. Pica and Long (1986) found interactional modifications more frequent
outside than inside the classroom, a finding that can also be explained by the
fact that roles are unequally distributed in the classroom, with the teacher as¬
signed many more discourse rights than the learners. Overall, these studies
suggest that when role relationships are asymmetrical, meaning negotiation is
inhibited. Other kinds of non-negotiated input may still be available in such
situations, however. Cathcart (1986), for instance, found that adult native
speakers were more likely to respond to requests for information from child
non-native speakers than were child native speakers.
A number of other factors have been found to influence negotiation: for
example, the nature of the task, the characteristics of the participants, and
participant structure. Much of the research that has investigated these factors
has involved classroom learners or has been conducted for a clear pedagogic
purpose (for example, to compare interaction in group and lockstep settings).
For this reason it will be considered in Chapter 13.
The presence of higher frequencies of discourse repair functions such as re¬
quests for clarification and confirmation has been taken as evidence that
higher levels of negotiation of meaning are occurring. This is, however,
262 External factors

questionable. Aston (1986) has pointed out that these discourse acts do not
unambiguously indicate negotiation of meaning, as the same procedures can
be used in non-problematic conversation. He argues that they are often used
‘to achieve a formal display of convergence of the participants’ worlds’ by
allowing them to perform ‘a ritual of understanding or agreement’. In other
words, negotiation can be motivated by the interactants’ need to display satis¬
factory outcomes rather than to overcome trouble sources.
In addition to repair work directed at solving problems of understanding,
there is also the repair of learners’ errors. Schwartz (1980) reports a general
preference for self-correction over other-correction in non-native speaker-
non-native speaker discourse. The same seems to be the case in native
speaker-non-native speaker discourse. Gaskill (1980) examined the types of
repair which an Iranian learner of English experienced in both elicited and
naturally-occurring conversations with native speakers. He found only
seventeen examples of other-correction in 50 pages of transcript. It would ap¬
pear, then, that native speakers typically ignore learners’ errors, a conclusion
supported by Chun, Day, Chenoweth, and Luppescu’s (1982) study of 28
ESL learners of mixed proficiency in Hawaii. In this case, less than 9 per cent
of the total errors were corrected. Also, it was ‘factual’ errors and ‘discourse
errors’ (for example, inappropriate openings, closings, and refusals) rather
than lexical or syntactic errors that were more likely to attract repair from
native speakers. Chun et al. suggest that the low level of repair reflects the
native speakers’ desire not to impair the cohesion of the discourse.
In the case of other-correction, a distinction can be made between on-
record and off-record feedback. Day, Chenoweth, Chun, and Luppescu
(1984) define the former as feedback which occurs when the native speaker
responds to the source of a learner’s language problems directly and unam¬
biguously, by means of a statement with declarative intonation. Off-record
repair is ambiguous and can have more than one interpretation; it can consist
of a question in the form of a confirmation check (although not all confirma¬
tion checks are corrective) or a statement. They found that adult ESL learners
received significantly more on-record than off-record feedback, perhaps
because as friends of the learners, the native speakers felt they could overtly
correct without threat to face.
In the case of linguistic modifications, there is substantial evidence that
their extent in native speaker speech varies according to the learners’ level of
development. Somewhat surprisingly, there have been few studies that have
investigated whether this is also true for interactional modifications. My
study (Ellis 1985d) suggests that some, but perhaps not all, of these modifica¬
tions are sensitive to the learner’s level. It was found that a teacher used signi¬
ficantly fewer self-repetitions, but more expansions, in interactions with two
learners when they had progressed beyond the stage of absolute beginners.
Also, the teacher switched from topics that required object identification to
topics that required the learners to make some kind of comment about an
 Input and interaction and second language acquisition 263

object. However, no change was observed in other interactional features such

as comprehension checks, requests for clarification and confirmation, types
of questions, and overall response to communication breakdown. Day et ah,
in the study referred to above, found that during game-like activities beginner
and intermediate learners received significantly more on-record repair than
did advanced learners. However, no differences were noted in conversations.
The effect of the learners’ proficiency on interactional modifications is clearly
an area in need of further study.
The study of interactional modifications in FT has blossomed and there is
now a rich literature to draw on. It is probably true to say, however, that the
research has focused on a rather narrow set of interactional phenomena
associated with the negotiation of meaning. The categories themselves are
less watertight than researchers sometimes admit—the function of a self- or
other-repetition, for example, is not always clear.8 Also, the research has been
somewhat restricted in the discourse types that it has studied—many of the
studies have examined data obtained in interview-type situations or by means
of information-gap activities.

4 Discourse structure

The study of interactional modifications involves examining the speech of

both participants in a conversation. As Long (1983a) has pointed out, acts
such as requests for clarification and confirmation take place within a
discourse context—they have life across utterances and speakers. However,
the study of individual discourse acts does not show how sequences of
discourse are constructed. For this it is necessary to examine the structure of
FT discourse.
Gass and Varonis (1985a; Varonis and Gass 1985) have developed a model
to describe the structure of ‘non-understanding routines’ where meaning
negotiation takes place. It consists of a ‘trigger’ (i.e. the utterance or part of
an utterance that creates a problem of understanding), an ‘indicator’ which
indicates that something in a previous utterance was not understood, a
‘response’ to the indicator, and finally a ‘reaction to the response’, which is
optional. The ‘indicator-response-reaction to the response’ portion of a non¬
understanding sequence is called a ‘pushdown’, because it has the effect of
pushing the conversation down rather than allowing it to proceed in a for¬
ward manner. The model is recursive in that it allows for the ‘response’ ele
ment itself to act as a ‘trigger’ for a further non-understanding routine. Table
7.4 provides a simple illustration of the model.
Other models exist. Ehrlich, Avery, and Yorio (1989) claim that the highly
specific nature of the problem-solving task they used yielded discourse with a
‘hierarchical constituent structure’. The task required the speaker (the native
speaker in native speaker-non-native speaker pairs) to give instructions
264 External factors

Utterance Function

NNS1: My father now is retire. Trigger

NNS2: Retire? Indicator
NNS1: Yes. Response
NNS2: Oh yeah. Reaction to Response

Table 7.4: A simple discourse model of the negotiation of meaning (example from
Gass and Varonis 1985a)

about how to fill in a matrix of 16 blank squares with different objects. All the
speakers proceeded through the task in the same manner beginning with an
‘orientation’, then moving on to the ‘identification’ of an object, before pro¬
viding a ‘description’ of it. The information provided in the ‘description’ was
also hierarchically organized, as some items of information depended on
others having been previously communicated. The model was used to distin¬
guish speakers who adopted a ‘skeletonizing’ strategy (i.e. provided only the
basic information needed to perform a task) from those who adopted an ‘em¬
broidering’ strategy (i.e. providing information that expands and embellishes
beyond what is required to perform the task).
The models that have been developed are essentially data-led models; that
is, they have been devised to describe the particular tokens of FT discourse eli¬
cited by the tasks used in the studies. Although there are now a number of
models of FT discourse structure currently available (such as those proposed
by Gass and Varonis and Ehrlich, Avery, and Yorio), it is not clear whether
these are sufficiently powerful to describe different discourse types and thus
to permit generalizations across studies. Nevertheless, the use of models that
account for discourse structure rather than taxonomies of discrete discourse
functions constitutes a definite advance, as it enables researchers to examine
the ‘pouring back and forth’ which Brown (1968:127) considers essential for
investigating how learners acquire language.

5 The functions of foreigner talk

Overall, three functions of foreigner talk can be identified: (1) to promote

communication, (2) to signal, implicitly or explicitly, speakers’ attitudes to¬
wards their interlocutors, and (3) to teach the target language implicitly.
Hatch (1983b) argues correctly that (1) is primary in that most adjustments
are geared to simplifying utterances to make them easier to process or to
clarify what has been said by either the native speaker or the non-native
speaker. Hatch characterizes (2) in terms of the special kind of affective bond
that FT can create between the native speaker and non-native speaker, but it
is also manifest in FT whose purpose is ‘talking down’ (Ferguson and Debose
 Input and interaction and second language acquisition 265

1977). In fact, it can reflect either downward divergence (such as when a

native speaker deliberately employs ungrammatical forms with a competent
native speaker to signal lack of respect), or downward convergence (such
as when a native speaker approximates the interlanguage forms used by the
native speaker as a way of signalling solidarity) . This double function of (2)
may help to explain why ungrammatical FT can occur between non-familiar
interlocutors in service or workplace encounters and between familiar inter¬
locutors in ordinary conversation. (3) is only ‘implicit’ because native
speakers do not usually have any pedagogic intent, although Naro (1983) in a
response to Hatch argues that FT can occur with an explicit teaching function
(i.e. with the intention of helping a learner learn).
Many of the formal characteristics of FT are very similar to those found in
other simplified registers such as learner language, caretaker talk, and
pidgins. This suggests that it reflects universal processes of simplification,
knowledge of which constitutes part of a speaker’s linguistic competence.
However, as Meisel (1980) has pointed out, there are differences between the
simplification found in FT and that observed in learner language. Whereas
both manifest restrictive simplification (for example, the use of an infinitive in
place of inflected verb forms), only learner language manifests elaborative
simplification (for example, the use of novel verb forms through processes
such as overgeneralization). Meisel suggests that restrictive simplification in
both registers serves ‘the purpose of achieving an optimal result in communi¬
cation’ (1980: 36), but elaborative simplification occurs when learners are
trying to complexify their interlanguage system.
An interesting question is how native speakers come to be able to adjust the
level of their FT to suit the level of individual learners. Hatch (1983b) con¬
siders three ways; (1) regression (native speakers move back through the
stages of development that characterized their own acquisition of language
until they find an appropriate level), (2) matching (native speakers assess a
learner’s current interlanguage state and then imitate the forms they observe
in it), and (3) negotiation (native speakers simplify and clarify in accordance
with the feedback they obtain from learners in communication with them).
The second is the explanation offered by Bloomfield (1933), but it seems un¬
likely, as it is probably asking too much of learners’ interlocutors to measure
simultaneously the learners’ phonology, lexicon, syntax, and discourse with
sufficient accuracy to adjust their own language output. The most likely ex¬
planation is (3), although (1) is also possible.

Interlanguage talk
Interlanguage talk (ILT) consists of the language that learners receive as input
when addressed by other learners. In Chapter 6 we noted that ILT constitutes
266 External factors

the primary source of input for many learners. The treatment will be brief
because there is a more extended discussion of the research in Chapter 13.
Two issues have figured prominently. The first concerns the extent to
which ILT provides learners with adequate access to the grammatical proper¬
ties of the target language. Not surprisingly, ILT has been found to be less
grammatical overall than FT or teacher talk (Pica and Doughty 1985a and
1985b; Wong-Fillmore 1992). Porter (1986) in a detailed study of the ILT
produced by intermediate and advanced L2 learners in pairwork and com¬
parable FT found that whereas only 6 per cent of FT was ‘faulty’, 20 per cent
of ILT proved to be so. Porter also found ILT to be sociolinguistically defi¬
cient. She looked at a number of speech acts such as expressing opinions,
agreement, and disagreement and found that the learners failed to use polite¬
ness strategies to the same extent as native speakers. In general they did not
generate the kind of sociocultural input needed for language learning. On the
plus side, however, both Pica and Doughty and Porter found that the learners
in their studies repeated only a very small amount of the faulty input they
heard.
The second major issue concerns whether ILT provides learners with the
same opportunities for negotiating meaning as occur in FT. Gass and Varonis
(1985a), in the study referred to above, reported an average of 0.50 ‘push¬
down routines’ in native speaker-native speaker pairs, 2.75 in native
speaker-non-native speaker pairs, and a massive 10.29 in non-native
speaker-non-native speaker pairs. Porter (1986) found that learners
prompted each other five times more than the native speakers prompted non¬
native speakers, while repair frequencies were similar. Overall, these studies
provide evidence that meaning negotiation is very extensive in ILT, more so
than in comparable FT discourse.
The quality of interlanguage talk is of considerable importance given the
current emphasis placed on small group work in communicative language
teaching (see Brumfit 1984). It is clear that ILT differs in some respects from
FT, but, if the theoretical arguments relating to the importance of meaning
negotiation for acquisition are accepted, a good psycholinguistic case can be
made for it (see Long and Porter 1985, and Chapter 13 of this book).

Summary

This section has examined three simple registers of language addressed to

learners. First, in studies of caretaker talk, it has been shown that the input
that children receive when learning their LI is well-formed and well-adjusted
linguistically to their level of development. Also, some caretakers seem adept
at helping children to establish and develop topics that they want to talk
about.
Foreigner talk resembles caretaker talk in some respects, but also differs
from it in others (for example, there are fewer yes/no questions). Both
 Input and interaction and second language acquisition 267

ungrammatical and grammatical FT occur, although it is not yet possible to

identify the precise social conditions that favour one over the other. In the
case of grammatical FT, three processes are evident: simplification, regular¬
ization, and elaboration. The modifications are continuous, influenced by the
learner’s stage of development and age. FT also displays interactional modi¬
fications directed at both managing and repairing discourse when either com¬
munication breakdown or learner error occurs. Figure 7.1 on page 258
summarizes the interactional modifications associated with these broad func¬
tions. Much of the research on FT discourse has been concerned with the
negotiation of meaning—the attempt to remedy mis- or non-understanding.
A number of models aimed at describing FT discourse have also been
developed, but no general model is available yet. Foreigner talk, like
caretaker talk, seems to be primarily motivated by the desire to achieve
communicative success.
Finally, we looked at interlanguage talk. This, not surprisingly, tends to be
less grammatical than FT, but it is characterized by more interactional modi¬
fications associated with the negotiation of meaning.

The effect of input and interaction on acquisition

We now turn to consider what effect input and interaction have on acquisi¬
tion. We will begin by considering LI acquisition research before moving on
to input studies in L2 acquisition.

Input and interaction in first language acquisition

A number of studies have investigated to what extent there is a relationship

between the language that caretakers address to children and acquisition. The
results have been somewhat contradictory, leading to controversy regarding
the role of input in LI acquisition. In an early study, Newport, Gleitman, and
Gleitman (1977) found little evidence of a close relationship between the fre¬
quency of specific linguistic features in mothers’ speech and the growth of the
same features in their children. They concluded that most aspects of language
structure in LI acquisition were insensitive to individual differences in the in¬
put, although the rate of growth of certain language-specific features (such as
auxiliary verbs and noun inflections) was sensitive. Furrow, Nelson, and
Benedict (1979) found much greater evidence of an effect for input. In this
study, four aspects of LI development (mean length of utterance, verbs per
utterance, noun phrases per utterance, and auxiliaries per verb phrase) were
related to a number of input measures. Barnes, Guttfreund, Satterly, and
Wells (1983) also produced evidence to suggest that input influenced lan¬
guage development. They reported significant correlations between the fre¬
quency of polar interrogatives and subject-verb inversion in the input, and
general semantic and syntactic development in the children and also between
268 External factors

intonation and auxiliary meanings and pragmatic functions. However, Gleit-

man, Newport, and Gleitman (1984), after reanalysing the data used in their
1977 study, reaffirmed their original results, finding few correlations between
input and language development measures.
As is often the case in acquisition research, these differences are difficult to
interpret or to reconcile. In part, they can be accounted for methodologically,
as Schwartz and Camarata (1975) and Bennett-Kastor (1988) point out. For
example, most of the studies used correlational statistics, which are difficult
to interpret because they do not provide any indication of the directionality of
a relationship. The studies also varied in whether and how they took account
of the children’s age and stage of development, both of which have been
shown to influence caretaker input. Snow (1986: 76) advances the argument
that ‘the task of learning a language is quite different in different stages of ac¬
quisition’ and that, consequently, it is reasonable to presume that different as¬
pects of caretaker talk may be important at different stages. It is worth noting
that all the studies report some relationships involving input and acquisition.
Fragile features 9 such as auxiliary verbs seem to be sensitive to input. Wells
(1985), for instance, found that caretakers increased the frequency of specific
auxiliary verbs in their input just before the same verbs appeared for the first
time in their children’s speech. The argument, therefore, is not whether input
has an effect, but rather what the extent of its effect is.
A somewhat more convincing case has been made out for the facilitative ef¬
fect of interactional features in caretaker talk. Wells (1985) has shown that
caretakers do not vary much in the extent to which they modify their speech
linguistically, but do vary considerably in the way they interact with their
children. This suggests that is the interactional rather than input features that
may be crucial in accounting for differences in children’s rate of acquisition.
One way in which interaction may help children learn language is by pro¬
viding them with opportunities to form vertical constructions. A vertical con¬
struction is built up gradually over several turns. Scollon (1976) provides a
number of examples of mother-child discourse, where the child produces a
meaningful statement over two or more turns:

Brenda: Hiding
Adult: What’s hiding?
Brenda: Balloon.

He suggests that vertical constructions prepare the child for the subsequent
production of horizontal constructions (the production of a meaningful state¬
ment within a single turn).
Other researchers point to the contribution that specific discourse acts
make to acquisition. Interaction rich in directives appears to foster rapid lan¬
guage acquisition, at least in the early stages (Ellis and Wells 1980), perhaps
because children are equipped with an ‘action strategy’ (Shatz 1978) that
makes them particularly receptive to directives, perhaps because directives
 Input and interaction and second language acquisition 269

are frequently related to the here-and-now and perhaps because they are
often linguistically very simple.
Expansions also appear to promote development. These are full, correct
versions of telegraphic child utterances, which often occur in the form of
acknowledgements or requests for confirmation:

Child: Pancakes away

Duh Duh stomach.
Mother: Pancakes away in the stomach, yes, that’s right.

Snow (1986) cites a number of studies that provide evidence of the positive
effect of expansions (for example, Malouf and Dodd 1972; Nelson 1977;
Cross 1978; Wells 1980), perhaps because they provide crucial bits of in¬
formation about syntax and morphology, although, as Snow notes, the same
information may also be made available to children in other kinds of input.
Children also seem to benefit from assistance in building conversations
about topics in which they are interested. Wells (1986a) and Snow (1986)
both emphasize the importance of ‘collaborative meaning making’ and dis¬
cuss some of the ways in which this is achieved. Successful conversations are
achieved when caretakers regularly check their understanding of what their
children have said, when they work hard to negotiate misunderstanding,
when they help to sustain and extend topics that their children have initiated,
when they give children opportunities to contribute to the conversation, no
matter how minimally, and when they are responsive to feedback cues. The
result is what Snow calls ‘semantically contingent speech’. However,
although the provision of such speech is considered to be advantageous for
language acquisition, it has not yet been shown that access to it is crucial for
success.

Input and interaction in second language acquisition

It is useful to distinguish four broad approaches in studies that have investig¬
ated the relationship between input/interaction and L2 acquisition. The first
simply seeks to establish whether the frequency of linguistic features in the in¬
put is related to the frequency of the same features in the learner language.
The second emphasizes the importance of input that is comprehensible to
learners. The third emphasizes the role of learner output in interaction. The
fourth looks more holistically at discourse by asking how the process of col¬
laborative discourse construction aids acquisition.

Input frequency and second language acquisition

The frequency hypothesis states that the order of L2 acquisition is determined
by the frequency with which different linguistic items occur in the input. The
hypothesis deals with the relationship between input and accuracy rather
270 External factors

than that between input and acquisition. The justification for considering it
here rests on the claim that the accuracy order mirrors the acquisition order
(see Chapter 3). The hypothesis was first advanced by Hatch and Wagner-
Gough (1976), who suggested that the limited range of topics about which
learners (particularly children) typically talk results in certain grammatical
features occurring with great frequency in the input. The more frequently oc¬
curring items, they claimed, were among those that emerged early in the
learners’ output. Input frequency, however, was only one of several explana¬
tions for the morpheme order which Hatch and Wagner-Gough considered.
Frequency, structural complexity, and cognitive learning difficulty overlap
and may together influence acquisition. Also, it should be remembered that
variations in the morpheme accuracy order were reported, reflecting the dif¬
ferent tasks learners were asked to perform (see the discussion on pages 90-6
in Chapter 3). This raises the thorny question of which order should be used
in correlational studies involving input frequency.
Table 7.5 summarizes some of the studies that have investigated the
input-accuracy-acquisition relationship. The limitations of these studies
need to be recognized. First, they are all correlational in nature, thus making
it impossible to make any statement about cause and effect. A statistically
significant correlation might simply indicate that learners and their interloc¬
utors have a similar communicative need to use the same grammatical
features. Second, some of the studies did not obtain measures of input
frequency from the same data sets used to examine learner accuracy (for
example, Larsen-Freeman 1976a; Long 1981b), thus making the correlations
they report additionally difficult to interpret. Third, as Snow and Hoefnagel-
Hohle point out, ‘it may be that quantity has a threshold rather than a graded
effect’ (1982: 424). If this is the case, the use of correlational statistics is
inappropriate, as these can only be used if the two sets are both evenly spread
throughout the total range. Fourth, it can be argued that it makes little sense
to investigate input-output relationships in data collected at the same time, as
any effect for frequency will not be immediately apparent. It is unfortunate
that there have been no L2 studies of the kind conducted in LI acquisition
where input frequencies were correlated with measures of language develop¬
ment (i.e. gains from one time to another). Only Lightbown’s (1983) study of
grade 6 ESL learners in Canada provides some indication of how input
available at one time can affect acquisition that becomes apparent at a later
time. In this study, the learners manifested over-use of \+-ing after having
received massive classroom exposure to this feature at the end of Grade 5.
Table 7.5 shows that the results have been very mixed. Whereas Larsen-
Freeman (1976a and 1976b), Lightbown (1980), Hamayan and Tucker
(1980), and Long (1981b) all found significant positive correlations between
input frequency and accuracy, Snow and Hoefnagel-Hohle (1982), Long and
Sato (1983), and Lightbown (1983) did not find any direct relationship. It is
perhaps premature on the basis of such mixed results to propose that ‘there
 Input and interaction and second language acquisition 271

Study Design Main results

Larsen-Freeman 1976a L2 accuracy orders of grammatical Spearman rank order correlations

morphemes correlated with input
frequency orders of same
morphemes in the classroom
speech of 2 ESL teachers.
Larsen-Freeman 1976b L2 accuracy orders of grammatical
morphemes correlated with
frequency order of same
morphemes in parental speech to
children (based on Brown 1973).
Flamayan and T ucker 1980 The accuracy order of 9 French
syntactic structures in speech
elicited from LI and L2 learners
correlated with frequency order of
same structures in teacher
classroom input.
Lightbown 1980 Order of appearance of L2 French
question words in speech of 2
children correlated with frequency
order of same question words in
speech addressed to the children by
a French research assistant.
Long 1981b Krashen’s average order of
acquisition of grammatical
morphemes correlated with the
frequency order of same
morphemes in the speech that NSs
addressed to 36 elementary
Japanese ESL students.
Snow and Hoefnagel- Achievement and gain scores of
Hohle 1982 American children acquiring L2
Dutch in Holland correlated with (1)
quantity of Dutch heard in the
classroom, (2) percentage of speech
directed specifically at learners, and
(3) amount of speech directed at
individual learners.
Long and Sato 1983 Krashen’s average order of
acquisition of grammatical
morphemes correlated with the
frequency order of the same
morphemes in ESL teachers’
speech.
Lightbown 1983 Grade 6 ESL students’ accuracy
order of -s morphemes in oral
communication activity correlated
with frequency order of same
features in samples of one teacher’s
speech and in textbook used by
students.
generally positive and significant.
Spearman rank order correlations
were positive and significant except
for imitation task.
Significant rank order correlations
reported for both LI and L2. The
frequency order reported for two
classrooms differed but was
significantly related to accuracy
orders in both cases.
Frequency with which question
words appeared in NS’s speech
matched closely the sequence of
question word development in L2
learners’ speech.
Spearman rank order correlation
positive and significant.
No significant relationships found.
No significant relationship found.
However, teachers’ frequency order
did correlate significantly with NSs’
frequency order, which in turn
correlated significantly with
Krashen’s average order.
No direct relationship between
accuracy and input frequencies
found, but some evidence that high
level of frequency of V + -ing in input
at earlier period (Grade 5) led to
over-use by the students.

Table 7.5: Input frequency-output accuracy studies in SLA research

exist preliminary data supporting a frequency effect’ (Larsen-Freeman and

Long 1991:133). It is possible that frequency may be more important at some
stages of acquisition (for example, elementary) than others, but no clear
272 External factors

conclusion is possible on the basis of these studies. All that can be said is that
the frequency hypothesis has not yet been properly tested.
Two other sets of studies, however, do lend support to the claim that input
frequency influences acquisition: (1) studies that have investigated the effects
of ungrammatical input, and (2) studies of the formulaic speech that many
learners produce.

Ungrammatical input

Evidence that ungrammatical input has a direct effect on acquisition comes

from a study by Gass and Lakshmanan (1991). They reanalysed data from
two of the learners investigated by Cazden et al. (1975)—Alberto and
Cheo—and found a striking correlation between the presence of subjectless
utterances in the input and in the production of the two learners. Again, the
correlations were derived from measures of input and output based on data
collected at the same time, but this study is more interesting than those re¬
viewed in Table 7.5 because of its longitudinal design. Gass and Lakshmanan
were able to show a strong and consistent relationship between input and
output over time (i.e. the presence of subjectless sentences in the input closely
matched those in the learners’ output from one time to the next). They were
also able to demonstrate that in the case of Alberto deviant input in a gram-
maticality judgement test was followed by a jump in this learner’s production
of subjectless sentences shortly afterwards. It should be noted, however, that
the learners’ LI (Spanish) permits subjectless sentences. It is possible, there¬
fore, that input and transfer work jointly to shape interlanguage develop¬
ment. As Gass and Lakshmanan put it, ‘the learner initially searches for
correspondences or matches in form between the native and the second
language’.10

Formulaic speech

The prevalence of formulaic speech in learner language was commented on in

Chapter 3. How do learners learn formulas? One possibility is that they re¬
spond to the high frequency of certain patterns and routines in the input. In
Ellis 1984b, for instance, I suggested that the formulas learnt by the three
classroom learners investigated reflected frequently occurring social and or¬
ganizational contexts that arose during the course of communicating in a
classroom environment. The prevalence of formulaic copula utterances with
subjects such as T, ‘this’, ‘that’, and ‘what’, which a number of researchers
have noted in learner language (Hakuta 1974; Rescorla and Okuda 1987;
Ellis 1988a), may also reflect input frequency. However, although it seems
reasonable to assume that frequency is a major determinant of the acquisi¬
tion of formulas, there is no research that clearly demonstrates that this is
so. Also, other factors are likely to be important—in particular, learners’
 Input and interaction and second language acquisition 273

communicative need to perform the language functions encoded by the

formulas with ease and fluency.
Overall, there is little evidence to support the claim that input frequency af¬
fects L2 acquisition, but there is also little evidence to refute it. Perhaps the
safest conclusion is that input frequency serves as one of the factors influenc¬
ing development, often combining with other factors such as LI transfer and
communicative need.

Comprehensible input and second language acquisition

A number of researchers see comprehensible input as a major causative factor
in L2 acquisition. The most influential theoretical positions are those ad¬
vanced by Krashen and Long.
Krashen’s Input Hypothesis (Krashen 1981; 1985; 1989) makes the fol¬
lowing claims:

1 Learners progress along the natural order by understanding input that

contains structures a little bit beyond their current level of competence.
2 Although comprehensible input is necessary for acquisition to take place, it
is not sufficient, as learners also need to be affectively disposed to ‘let in’ the
input they comprehend.
3 Input becomes comprehensible as a result of simplification and with the
help of contextual and extralinguistic clues; ‘fine-tuning’ (i.e. ensuring that
learners receive input rich in the specific linguistic property they are due to
acquire next) is not necessary.
4 Speaking is the result of acquisition, not its cause; learner production does
not contribute directly to acquisition.11

Long differs from Krashen primarily with regard to (3). While acknowl¬
edging that simplified input and context can play a role in making input com
prehensible, Long stresses the importance of the interactional modifications
that occur in negotiating meaning when a communication problem arises. In
other words, Long’s argument is that interactive input is more important than
non-interactive input.
These theoretical claims have led to research into two aspects of input: (1)
the relationship between input (both interactive and non-interactive) and
comprehension, and (2) the relationship between linguistic/conversational
adjustments and acquisition. To date, there has been far more research dir¬
ected at (1), mainly because of the difficulties of designing studies to investi¬
gate (2). In recognition of these difficulties, Long (1985a: 378) suggested the
following three steps as a way of gaining insight into how input/interaction
affects acquisition:
Step 1: Show that (a) linguistic/conversational adjustments promote (b)
comprehension of input.
Step 2: Show that (b) comprehensible input promotes (c) acquisition.
274 External factors

Step 3: Deduce that (a) linguistic/conversational adjustments promote (c)

acquisition.

Young (1988b) points out that although a number of studies have demon¬
strated step (1) there have been few which have tackled (2), which is, in fact,
the crucial step.

1 The relationship between interactive and non-interactive input and

comprehension

What evidence is there that linguistic/conversational adjustments promote

comprehension of input, and do some kinds of adjustments work better than
others? To answer these questions, it is helpful to separate out research which
has investigated non-interactive input from that into interactive input.
One type of non-interactive adjustment concerns speech rate. A number of
studies (for example, Long 1985a; Kelch 1985; Mannon 1986; Fujimoto,
Lubin, Sasaki, and Long 1986) provide evidence to suggest that a slower rate
aids comprehension, but in many cases speech rate was investigated alongside
other variables, making it difficult to assess the effect of speech rate per se.
Other methodological problems also make interpretation of the results diffi¬
cult. For example, Kelch (1985) claimed to present dictation passages at two
different speeds (191 and 124 words per minute) but in fact allowed 45
seconds between each sentence dictated, thus introducing another variable
(pause duration) into the study.
However, two other, better-designed studies indicate that rate of speech is
a significant factor. Conrad (1989) asked 29 native speakers, 17 high-level
non-native speakers, and 11 medium-level non-native speakers to recall five
time-compressed recordings of 16 English sentences. Each sentence was pre¬
sented five times at different speeds: 450, 320, 253, 216, and 196 words per
minute. The native speakers managed nearly complete recall of all the sen¬
tences by the second trial (320 words per minute), but the non-native speak¬
ers of both proficiency levels experienced considerable difficulty even with the
slowest rate after the fifth hearing. Griffiths (1990) investigated the effects of
varying the speech rate in three 350-400 word passages on the comprehen¬
sion of 15 lower-intermediate level adult non-native speakers. The speeds
varied from 94-107 wpm (slow) to 143-54 (medium) and 191-206 (fast).
The subjects manifested significantly reduced comprehension, measured by
means of true/false questions, at fast rates, but there was no difference be¬
tween the ‘normal’ and ‘slow’ rates. These studies suggest that there may be a
threshold level - around 200 wpm - below which intermediate and advanced
level learners experience little difficulty in comprehending and above which
they might.
Of particular interest to researchers is the nature of the modifications that
are most effective in promoting comprehension. Parker and Chaudron
(1987) review 12 experimental studies (including Johnson 1981; Blau 1982;
 Input and interaction and second language acquisition 275

Cervantes 1983; R. Brown 1987; Chaudron and Richards 1986) of the effects
of input modifications on comprehension and conclude that although lin¬
guistic modifications (for example, simpler syntax and vocabulary) helped
comprehension they did not do so consistently. In contrast, what they call
‘elaborative modifications’ had a consistent effect on comprehension. They
distinguish two types of elaborative modifications: those that contribute to
redundancy (such as repetition of constituents, paraphrase, use of synonyms,
use of left dislocation, and slower speech), and those that help to make the
thematic structure explicit (such as extraposition and cleft constructions).
However, Parker and Chaudron’s own study of 43 undergraduate and
graduate ESL learners, who read one of two reading passages which differed
according to whether they had been modified by incorporating linguistic or
elaborative modifications, failed to find any significant differences in compre¬
hension. They explain this unexpected result by suggesting that the overall
lexical and syntactic difficulty of both passages may have been so great as to
negate the effects of the elaborative modifications. If this is right, it suggests
that elaborative modifications will only benefit comprehension if the level of
linguistic difficulty of the input does not exceed a certain threshold.
Issidorides and Hulstijn (1992) also failed to find that linguistic modifica¬
tions had any effect on comprehension. This study examined the effects of
simplifying Dutch AdvVSO sentences by replacing them with the ungram¬
matical AdvSVO and AdvSOV sentences. These ‘simplifications’ were motiv¬
ated by the fact that learners of L2 Dutch and L2 German have been found to
experience production difficulties with verb-subject inversion after initial ad¬
verbs (see Chapter 3). In an experimental study that required three groups of
subjects (Dutch native speakers, English and Turkish learners of L2 Dutch) to
state what the subject or agent of a series of sentences was, AdvVSO sentences
were found to be no more difficult to understand than modified, ungrammat¬
ical sentences (providing that the messages conveyed by the sentences were se¬
mantically plausible). Issidorides and Hulstijn conclude that ‘the fact that
non-native speakers have difficulties in producing a certain grammatical
structure ... does not imply that such a structure is also more difficult to un¬
derstand in the speech of others’ (1992: 167).
Another question of interest concerns the role of the learners’ LI in L2
comprehension. Holobrow, Lambert, and Sayegh (1984) found that 10-year-
old Canadian children comprehended better when they heard an oral text in
their LI and at the same time read an L2 version of the text than when they
simultaneously heard and read texts in the L2 or when they just read L2 texts.
However, when Hawkins (1988) replicated this study with British under¬
graduate students of L2 Lrench, he found that the bimodal condition (i.e. lis¬
tening to and reading texts in the L2) worked best. No conclusions are
possible on the basis of these studies, but they raise the issue of the role of the
LI in helping to make input comprehensible.
276 External factors

In the case of interactive discourse, two variables have been found to influ¬
ence comprehension: the amount and type of information, and the extent to
which the participants engage in negotiation of meaning. Derwing (1989)
found that some native speakers attempted to include much more informa¬
tion than they used with native speaker interlocutors about a film they had
seen, with the result that their non-native speaker interlocutors failed to un¬
derstand their narratives. The native speakers who were successful in com¬
municating did not differ in the kinds of information (crucial, major, and
minor) that they used when talking to native speakers and non-native speak¬
ers. This study suggests that redundancy involving increased use of back¬
ground detail is not helpful to comprehension, although, as Derwing notes,
this does not mean that other kinds of redundancy (such as repetition) are not
helpful. Ehrlich, Avery, and Yorio (1989) also report that their ‘em¬
broiderers’ created problems for the non-native speakers by making it diffi¬
cult for them to identify essential information and the source of
communication problems.
Comprehension also appears to benefit from opportunities for negotiation
of meaning. Pica, Young, and Doughty (1987) compared the effect of three
types of input on the ability of sixteen low-intermediate ESL learners to com¬
prehend oral instructions. The three types were (1) unmodified input (i.e. in¬
put of the kind that native speakers use when addressing each other), (2)
premodified input (i.e. input that had been simplified and made more redund¬
ant), and (3) interactionally modified input (i.e. the subjects listened to
unmodified instructions but were given the opportunity to seek clarification).
The results showed that (3) resulted in the highest levels of comprehension.
Subsequent studies by Loschky (1989), Tanaka (1991), and Yamazaki
(1991) confirm these results. However, it should be noted that in all these stu¬
dies the opportunity for negotiation led to considerable repetition and
rephrasings with the result that (3) provided the learners with much more in¬
put than was available in (1) and (2). It is not clear, therefore, whether the ad¬
vantage found for the interactionally modified input arises from greater
quantity of input or better quality (i.e. input made relevant through the nego¬
tiation of meaning). A subsequent study by Pica (1992) found that when the
extent of the input adjustments was carefully controlled in the premodified
and interactionally modified conditions no difference in comprehension was
found,12 although a post-hoc analysis suggested that those learners rated as
having lower comprehension ability by their teachers benefited from the op¬
portunity to interact. Pica’s study also found no significant differences in the
comprehension of learners who observed negotiation taking place but did not
actually negotiate themselves, and those who actively participated.
To sum up, there is mixed evidence regarding the value of linguistically
simplified input for promoting comprehension. Whereas speech rate does
have a clear effect, grammatical modifications do not always result in im¬
proved comprehension. Firmer support exists for the beneficial effect of inter-
 Input and interaction and second language acquisition 111

actionally modified input on comprehension. We should note, however, that

the presence of interactional modification is no guarantee that comprehen¬
sion has taken place. As both Aston (1986) and Hawkins (1985) point out,
learners may choose to feign comprehension after negotiation rather than
continue to demonstrate to their interlocutors that they have not understood.
We will now consider to what extent comprehensible input promotes acquisi¬
tion (step (2) in Long’s suggested procedure).

Source of evidence Brief explanation

Caretaker speech Caretaker speech to young children is roughly tuned

Foreigner talk
Silent period
Age difference
Comparative method studies
Immersion programmes
Bilingual programmes
Delayed LI and L2 acquisition
Reading and vocabulary acquisition
Reading and spelling acquisition
to the children’s receptive abilities and is motivated
by the need to aid comprehension.
Foreigner talk to NNSs is also roughly tuned and
functions as an aid to comprehension.
Some young children go through a silent period in L2
acquisition. During this period they do not produce
but nevertheless learn the L2.
Krashen (1985:12) argues that ‘older learners obtain
more comprehensible input’ and that this may explain
why they learn more quickly initially than younger
learners.
The studies show that methods that supply plenty of
comprehensible input (e.g. Total Physical Response)
are more successful than methods that supply little
(e.g. audiolingualism).
Immersion programmes have generally been found
superior to foreign/second language programmes -
again because they supply plenty of comprehensible
input.
The success of different kinds of bilingual
programmes is related to the extent to which they
supply comprehensible input.
Studies of children in both LI and L2 acquisition who
are deprived of comprehensible input (e.g. because
their parents are deaf) show that acquisition is
delayed or non-existent.
Studies indicate that children are able to increase
their LI vocabulary and also develop a ‘deep’
understanding of new words through pleasure
reading.
Studies indicate that spelling can be most effectively
acquired through exposure to the written word in
extensive reading undertaken for pleasure.

Table 7.6: Indirect evidence in support of the input hypothesis

278 External factors

2 The relationship between comprehensible input and acquisition

The evidence is primarily indirect. Together Long (1983b), Larsen-Freeman

and Long (1991), and Krashen (1985; 1989) cite literature relating to ten
sources that they claim support the claim that comprehensible input contrib¬
utes to acquisition. These sources are described in Table 7.6.
Overall, this evidence supports the commonsense assumption that learners
need to understand input in order to learn from it. It is less clear, however,
whether this evidence warrants the claim that comprehensible input is a caus¬
ative factor in L2 acquisition. A careful examination of each source of evid¬
ence reveals various doubts. In many cases, the evidence is correlational in
nature: for example, caretaker and foreigner talk co-occur with successful ac¬
quisition; extensive reading is associated with vocabulary development and
improved spelling; older learners learn faster and engage in more meaning ne¬
gotiation. Therefore, it cannot be said to demonstrate that comprehensible
input causes acquisition. In other cases, the validity of the evidence is in
doubt. For example, the comparative method studies have problematic de¬
signs and have produced very mixed results; and many learners do not go
through an extended silent period (see Chapter 3). In yet other cases, the
available evidence lends less than complete support (for example, learners in
long-term immersion programmes do not develop full target-language profi¬
ciency). Long (1983b) considers the fact that deprivation of comprehensible
input results in delayed acquisition the strongest evidence, but Larsen-
Freeman (1983) argues quite convincingly that learners can assimilate useful
information about an L2 without understanding input (for example, in¬
formation relating to the phonology of an L2). She also makes the point that
learners can work by themselves on unmodified input—such as that found in
TV programmes—and so gain input that helps them learn. Thus, at least
some L2 learning may take place without comprehensible input and, more
importantly, comprehending input may not result in acquisition.
There have been a number of critiques of the input hypothesis (Gregg 1984;
Faerch and Kasper 1986a; Sharwood Smith 1986; White 1987a; Gass 1988;
ElliS 1990a and 1991a). Gass (1988) makes the point that it is not compre¬
hensible input but rather comprehended input that is important, but the re¬
search reported above does not provide an account of how comprehending
input is supposed to result in acquisition. In fact, little attention has been paid
to what ‘comprehension’ entails. Rost defines listening comprehension as ‘es¬
sentially an inferential process based on the perception of cues rather than
straightforward matching of sound to meaning’ (1990: 33). This suggests
that understanding does not necessitate close attention to linguistic form.
Faerch and Kasper (1986) also recognize the importance of ‘top-down pro¬
cesses’, in which learners utilize contextual information and existing know¬
ledge to understand what is said, but they also point out that they may
sometimes make use of ‘bottom-up processes’, where they pay closer
 Input and interaction and second language acquisition 279

attention to the linguistic forms in the message. The question that remains
unanswered, then, is what type of comprehension processes are needed for
acquisition to take place. If learners can rely extensively on top-down pro¬
cessing they may pay little attention to the form of the input and may
therefore not acquire anything new. Fserch and Kasper argue that only when
there is a ‘gap’ between the input and the learner’s current interlanguage and,
crucially, when the learner perceives the gap as a gap in knowledge, will ac¬
quisition take place. Sharwood Smith (1986) argues similarly that the pro¬
cesses of comprehension and acquisition are not the same, and suggests that
input has a ‘dual relevance’—there is input that helps learners to interpret
meaning, and there is input that learners use to advance their interlanguages.
A second major criticism focuses on the claim that comprehensible input is
necessary for acquisition.13 White (1987a) has argued that a considerable
part of acquisition is ‘input-free’. As we will see in Chapter 10, she claims that
certain types of overgeneralizations which learners make cannot be unlearnt
simply by understanding input. They require negative evidence (for example,
in the form of corrective feedback) which in naturalistic acquisition may not
be available to the learner. She also claims that learners are able to go beyond
the evidence available in the input, and develop knowledge of target-language
rules by projecting from their existing knowledge (a point also discussed
more fully in Chapter 10). Finally, she argues that in the case of some struc¬
tures (for example, English passive constructions), it may be the failure to un¬
derstand input that leads to learning. As she puts it ‘the driving force for
grammar change is that input is incomprehensible, rather than comprehens¬
ible ...’ (1987a: 95). White’s idea is that failure to understand a sentence may
force the learner to pay closer attention to its syntactical properties in order to
obtain clues about its meaning and, as such, reflects the views of Fserch and
Kasper (1986) and Sharwood Smith (1986) about comprehension and learn¬
ing discussed above.
These criticisms can be accommodated if the hypothesis is modified in the
following way:
Comprehensible input can facilitate acquisition but (1) is not a necessary
condition of acquisition, and (2) does not guarantee that acquisition will
take place.
The question arises as to whether such a modified hypothesis has much
explanatory value. Arguably, what is needed is a hypothesis that explains
how learners’ attempts to comprehend input can lead to acquisition.

3 The relationship between input/interactional modifications and

acquisition

Finally, we will consider the few studies which have attempted to establish a
more direct relationship between input/interactional modifications and ac¬
quisition. Pica (1992) illustrates how the negotiation of meaning provides
280 External factors

learners with information about the semantic and structural properties of the
target language (English). Native speakers (NS) responded to non-native
speakers’ (NNS) triggers by modifying their utterances semantically and/or
formally through the segmentation and movement of input constituents, as in
this example:

NNS: Ok, you have a house which has third floor.

NS: Three floors right.
NNS: Three floors.

However, as Pica admits, it is not clear from such data whether learners use
the information supplied by such exchanges to adjust their interlanguage
systems.
A number of experimental studies, however, have been carried out to dis¬
cover whether negotiation leads to interlanguage development. Although the
results of these studies have been mixed, they give some support to the claim
that FT modifications help acquisition, at least where vocabulary is con¬
cerned. The studies are summarized in Table 7.7.
These studies give support to two general conclusions. First, Fi’s (1989)
study suggests that providing learners with contextual cues that help them to
understand the meanings of words results not only in better comprehension
but also in better retention of the words. Second, Tanaka’s (1991) and Yama-
zaki’s (1991) studies indicate that providing learners with opportunities to
modify input interactionally improves comprehension and enables them to
learn more new words than simply providing them with unmodified or pre¬
modified input. These findings augur well for Fong’s claims regarding the im¬
portance of interactionally adjusted input. In contrast, Foschky’s (1989)
study failed to find any advantage for interactionally-modified input over
baseline or presimplified input where vocabulary retention was concerned. It
should be noted, however, that this study, unlike the studies by Tanaka and
Yamazaki, had a grammatical focus (locative particles in Japanese), so it is
possible that the learners’ attention was directed at this rather than at the new
vocabulary items. To date there has been no empirical test of the claim that
negdtiation of meaning aids the acquisition of new grammatical features.

Learner output and acquisition

Interaction also provides learners with the opportunity to talk in the F2. Ac¬
cording to Krashen (1985), this has no direct effect on acquisition. However,
other researchers have argued differently, viewing learner output as contri¬
buting to interlanguage development.
Following Krashen (1989), two different hypotheses that allocate a role to
output can be identified. The skill-building hypothesis states that we first
learn rules or items consciously and then gradually automatize them through
practice. We will consider this when we examine cognitive theories of F2
 Input and interaction and second language acquisition 281

Study Design Main results

Li (1989) Advanced Chinese ESL learners Cue-adequate sentences were

exposed to sentences under two
conditions: (1) cue-adequate
(meaning made clear by context)
and (2) cue-inadequate.
Loschky (1989) Three groups of learners of
Japanese as a second language
were given lessons on locative
expressions under three different
conditions: (1) baseline input, (2)
premodified input, and (3)
interactional^ modified input.
Tanaka (1991) Three groups of Japanese high
school ESL students were asked
to respond to instructions given
under three conditions: (1)
baseline input, (2) pre-modified
input, and (3) interactional^
modified input. Instructions
contained a number of new
words. Pre-, post- and follow-up
vocabulary tests were
administered.
Yamazaki (1991) Same design as for Tanaka’s
study.
better understood and led to better
retention of difficult words.
Learners receiving interactional^
modified input had higher
comprehension scores than learners
receiving premodified input, who
had higher scores than learners
receiving baseline input. No
differences in the three groups for
retention of the new vocabulary
items introduced in the lessons.
Interactional^ modified input
resulted in better comprehension of
the instructions. It also resulted in
more words being learnt and
retained overtime.
Interactional^ modified input
resulted in better comprehension
and also in more words being learnt
(as evident in post-test scores), but
learning advantage was not retained
over time (as evident in follow-up
test scores).14

Table 7.7: Studies investigating the direct effects of input/interaction modifications

on L2 acquisition

acquisition in Chapter 9 and the utility of formal practice in Chapter 14. Here
we will focus on the second hypothesis Krashen considers, the output
hypothesis.
This comes in two forms, according to Krashen. First, there is ‘output plus
correction’. According to this, learners try out rules or items in production
and then use the corrections they receive from other speakers to confirm or
disconfirm them. Schachter (1986b) points out that metalinguistic informa¬
tion relating to the correctness of learners’ production is available both dir¬
ectly (through corrections) and indirectly (through confirmation checks,
clarification requests, and failure to understand).
282 External factors

The second form of the output hypothesis involves the idea of comprehens¬
ible output. Swain (1985) argues that learners need the opportunity for mean¬
ingful use of their linguistic resources to achieve full grammatical
competence. She argues that when learners experience communicative fail¬
ure, they are pushed into making their output more precise, coherent, and ap¬
propriate. She also argues that production may encourage learners to move
from semantic (top-down) to syntactic (bottom-up) processing. Whereas
comprehension of a message can take place with little syntactic analysis of the
input, production forces learners to pay attention to the means of expression.
The evidence indicating that comprehensible output is important for ac¬
quisition is largely indirect. A number of studies (Harley and Swain 1978;
Harley 1988; Harley, Allen, Cummins, and Swain 1990) have shown that al¬
though immersion learners achieve considerable confidence in using the L2
and considerable discourse skills, they fail to develop more marked grammat¬
ical distinctions, such as that between French passe compose and imparfait,
and full sociolinguistic competence. Swam argues that this be cannot be ex¬
plained by a lack of comprehensible input, as immersion classrooms are rich
in this. She speculates that it might be because the learners had limited oppor¬
tunity to talk in the classroom and were not ‘pushed’ in the output they did
produce. Allen, Swain, Harley, and Cummins (1990) conducted an observa¬
tional study of interaction in French immersion classrooms in nine grade 3
and ten grade 6 classrooms and found that the students’ responses were typic¬
ally ‘minimal’. Less than 15 per cent of students’ utterances in French were
more than a clause in length. Also, there was little evidence of any systematic
correction of the errors learners made in their output. Only 19 per cent of
overall grammatical errors were corrected and often in ‘a confusing and
unsystematic way’. They conclude that ‘there was little indication that stu¬
dents were being pushed towards a more coherent and accurate use of the tar¬
get language’ (1990: 67).
It is important to recognize that Swain’s claim is that production will aid
acquisition only when the learner is pushed. Thus plentiful opportunities to
speak, such as those that Wes undoubtedly had in Hawaii (see Schmidt 1983
and Chapter 6 of this book), will not in themselves guarantee acquisition.
Learners like Wes who develop high levels of strategic competence may be
able to communicate efficiently without much grammatical competence, and
so may never experience the need to improve their output in order to make
themselves understood.
Both versions of the output hypothesis attribute considerable importance
to feedback, both direct and indirect. In the case of ‘output plus correction’,
feedback is necessary to supply learners with metalinguistic information,
while in the case of ‘comprehensible output’ it is necessary to push learners to
improve the accuracy of their production in order to make themselves un¬
derstood. We have already seen that foreigner talk provides learners with
little direct feedback1’ but with plenty of indirect feedback in the form of
 Input and interaction and second language acquisition 283

clarification requests, confirmation requests, etc. It would seem, therefore,

that it is indirect feedback that is crucial. Three questions arise: (1) Does
indirect feedback actually lead to more accurate learner output? (2) Do
certain types of indirect feedback work better than others in this respect? and
(3) Do learners actually learn as a result of pushed output?
A series of studies by Pica have addressed the first two questions. Pica (1988)
examined the interactions between a native speaker and ten non-native
speakers of English in order to discover whether indirect feedback led to im¬
proved output. She found that the learners did produce more grammatical
output when the native speakers requested confirmation or clarification, but
in less than half of their total responses. In a subsequent study, however, Pica,
Holliday, Lewis, and Morgenthaler (1989) found that the crucial factor was
the nature of the feedback signals the native speaker provided. The learners
(10 Japanese adults) were much more likely to produce output modifications
in response to clarification requests than to confirmation requests and repeti¬
tions. This was because clarification requests constitute an ‘open signal’, leav¬
ing it up to the learners how to resolve the comprehension problems, whereas
confirmation requests and repetitions, where the native speaker models what
the learner intended to mean, remove the need for improved output. Interest¬
ingly, this study also showed that many of the modifications that learners
made were grammatical and not just semantic in nature.
Evidence that learners improve the grammaticality of their utterances
when pushed does not of course constitute evidence that acquisition takes
place. In an exploratory study involving just six learners (three experimental
and three control), Nobuyoshi and Ellis (1993) sought to establish whether
pushed output resulted in improved performance over time. In two-way in¬
formation gap tasks, which each learner performed with the researcher, the
experimental subjects experienced ‘focused meaning negotiation’, where they
received a clarification request every time they produced an utterance con¬
taining a past tense error.16 The control subjects experienced ‘unfocused
meaning negotiation’ (i.e. negotiation only took place when there was a genu¬
ine communication problem). Both sets of learners believed they were com¬
municating rather than practising language. Nobuyoshi repeated the
information-gap task with both groups one week later, this time without a
special focus on past tense. Two out of three experimental learners improved
the accuracy of their use of the past tense as a result of the requests for clari¬
fication and maintained this improvement one week later. The control sub¬
jects showed no improvement on either occasion. This study intriguingly
suggests that pushing learners to produce more comprehensible output may
have a long-term effect, but not necessarily for all learners.17
There is also another sense in which learners may benefit from being
‘pushed’. The studies reported above examined the effect of clarification re¬
quests on the subsequent accuracy of learners’ output. Some learners, how¬
ever, may have achieved a high level of accuracy but only with a restricted set
284 External factors

of ‘simple’ constructions. Such learners need to experience situations that call

for more complex output, as, for example, when they are called on to demon¬
strate mastery of the forms required to communicate in a sociolinguistically
appropriate manner with different addressees (see Chapter 5).
So far, there is little hard evidence to support the output hypothesis, al¬
though there is sufficient to suggest that it is worth pursuing. It is possible that
when learners produce pushed output they are able to develop greater control
over the features they have already acquired, as demonstrated in Nobuyoshi
and Ellis’s study. It is not clear whether pushed output can result in the ac¬
quisition of new linguistic features.

Collaborative discourse and second language acquisition

We saw that LI researchers have produced evidence that suggests that chil¬
dren may acquire syntactical structures as a result of learning how to particip¬
ate in conversations. Similar claims have been made by L2 researchers.
Hatch, for instance, proposes:
One learns how to do conversations, one learns how to interact verbally,
and out of this interaction syntactic structures are developed (1978b: 404).
Researchers have documented a number of ways in which this might happen.
One way is through the use of an incorporation strategy. Wagner-Gough
(1975) investigated the conversations which an Iranian boy, Homer, had
with native speakers and showed how he put together utterances by ‘borrow¬
ing’ a chunk from the preceding discourse and then extending it by affixing an
element to the beginning or end, as in these examples:
1 Mark: Come here.
Homer: No come here.
2 Judy: Where’s Mark?
Homer: Where’s Mark is school. (= Mark is at school.)

This strategy may explain why ‘no + verb’ constructions are so common in
early L2 acquisition (see Chapter 3), although other explanations also exist.
Peck (1978) gives examples of a related but slightly different strategy which
she calls ‘functions’ (a term borrowed from Keenan’s (1974) study of LI ac¬
quisition). These occurred frequently in child-child discourse involving
Angel, a 7-year-old Mexican boy. Lunctions involve repeating with or with¬
out some modification part or all of the previous utterance. There are differ¬
ent types. A ‘substitution function’, for example, involves replacing a
constituent in the previous utterance, as in this example:
Joe: You know what?
Angel: You know why?

Peck suggests that this kind of discourse enables learners to practise syntax
and pronunciation.
 Input and interaction and second language acquisition 285

Vertical constructions of the kind reported by Scollon (1976) in LI acquisi¬

tion have also been found in L2 acquisition. Hatch (1978b: 407) provides this
example in an interaction involving Paul, Huang’s (1970) subject:

Paul: Oh-oh!
J: What?
Paul: This (points at an ant)
J: It’s an ant.
Paul: Ant.

The construction that Paul builds over three utterances is ‘Oh-oh, this ant’.
Of course, these studies are not able to show that ‘scaffolding’ of the kind
illustrated in these examples results in acquisition as opposed to just aiding
production. In Ellis (1985d), however, I found evidence to suggest that this
might indeed be the case. Interactions between two learners and a teacher
were examined in order to identify learner utterances which featured ‘new’
syntactic patterns (for example, the first occurrence of two constituent utter¬
ances consisting of noun + noun). I then studied what was going on inter-
actionally that might have helped the learner to produce them. It was found
that ‘new’ structures emerged when the learners were allowed to initiate a
topic (if necessary by the teacher abandoning his own topic) and when the
teacher helped the learners by supplying crucial chunks of language at the
right moment.
Collaborative discourse may also work against acquisition, however. A
number of studies suggest that learners can sometimes exploit discourse
strategies to compensate for morphological features they have not acquired.
Studies of L2 German by Dittmar (1981), Klein (1981), and Meisel (1987)
and of L2 French by Trevise and Porquier (1986) show how learners are able
to refer to time and space without recourse to target-language inflections.
They rely instead on adverbials, implicit reference, order of mention, and
interlocutor scaffolding. This research relates to the question of how learners
move from the pragmatic to the grammatical mode (how they grammaticalize
their speech) and is considered in greater detail in Chapter 9. What is import¬
ant here is to note that collaborative discourse may be so successful that it
inhibits grammatical acquisition.
It would seem, then, that collaborative discourse can both facilitate the ac¬
quisition of L2 syntax and restrict it. The contradiction may be more appar¬
ent than real, however. Where basic syntax is concerned, it appears to help,
particularly in the early stages of acquisition. Where morphological features
such as tense markers are concerned, it may obviate the need for
development.
286 External factors

Summary

In this section we have looked at some of the research that has investigated the
relationship between input/interaction and acquisition.
First language acquisition research has shown that the frequency of lin¬
guistic forms in caretakers’ speech is associated with developmental gains in
children, particularly where fragile features (for example, auxiliary verbs) are
concerned. Flowever, the claim that input frequency determines acquisition is
controversial. More important for rapid language development are the inter¬
actional features of caretaker talk (for example, directives, expansions, ‘col¬
laborative meaning making’, and the negotiation of misunderstandings).
The relationship between input/interaction and second language acquisi¬
tion has been examined in four major ways:

1 Input frequency-output accuracy studies

These have produced mixed results and suffer from methodological prob¬
lems, but the positive correlations between the frequency in input of specific
grammatical morphemes reported by a number of studies have received con¬
firmation from studies of the effects of ungrammatical input on learner
output.

2 Comprehensible input studies

The input hypothesis and Long’s claims about the role of interaction have
motivated a substantial amount of research, although they have also been
subject to considerable criticism on the grounds that they fail to specify how
comprehending input leads to acquisition and that they overstate the import¬
ance of comprehensible input by claiming that it is necessary for acquisition.
Much of the research has concentrated on identifying the features that help to
make non-interactive and interactive input comprehensible. In the case of
non-interactive input, speech rate, elaborative modifications, and bimodal in¬
put (i.e. both oral and written) have been found to aid comprehension. In the
case of interactive input, the amount of information and the extent to which
meaning is negotiated through interactional adjustments have been shown to
be significant variables. There is substantial indirect evidence linking compre¬
hensible input to acquisition, but much of it is controversial. There is little in
the way of direct evidence and what there is relates to the acquisition of
vocabulary rather than syntax or morphology.

3 Learner output and acquisition

Swain has argued that comprehensible output contributes to acquisition in

that learners need to be pushed into producing output that is concise, coher¬
ent, and appropriate in order to develop full grammatical competence. There
 Input and interaction and second language acquisition 287

is evidence to show that indirect feedback in the form of clarification requests

pushes learners to improve their output, but little evidence as yet to show that
this then leads to acquisition.

4 Collaborative discourse and acquisition

Illustrative evidence has been provided to show that the joint efforts of native
speakers and non-native speakers to construct discourse promotes acquisi¬
tion in a number of ways: for example, through the use of an incorporation
strategy (the construction of utterances by borrowing from and extending
elements from the preceding discourse), functions (repeating with or without
modification some previous utterance), and vertical constructions (building
up an utterance over several turns). However, collaborative discourse may
remove the need for learners to acquire certain morphological features.

Conclusion

Young (1988b: 128) concludes his own survey of input and interaction re¬
search with the comment that ‘there is still a great deal of beating about the
bush’, by which he means that relatively few studies have tried to investigate
to what extent and in what ways input and interaction influence acquisition.
One reason for this is the methodological problems that input researchers
face. In LI acquisition research it is not clear how potentially confounding
variables such as the child’s age and stage of development can be controlled so
as to investigate the effects of input on acquisition. In L2 acquisition research,
this problem is exacerbated by the difficulty of obtaining representative
samples of the input that individual learners receive, for where the LI learner
is situationally restricted to interacting with a few identifiable interlocutors,
the L2 learner, especially when an adult, is not. Second, whereas a number of
input studies on LI acquisition have been longitudinal, nearly all those into
L2 acquisition have been cross-sectional. Third, there has been an over-
reliance on correlational studies at the expense of experimental studies, mak¬
ing it difficult to determine what is cause and what is effect. Progress requires
studies that (1) reliably sample the input data, (2) are longitudinal, and (3) are
experimental.
Despite these methodological problems, a number of insights regarding the
role of input and interaction have been gained. We have some understanding
of how interactional modifications affect the comprehensibility of texts. It is
also becoming apparent that different kinds of input and interaction are
needed to facilitate acquisition at different stages of learners’ development,
and that input and interaction may or may not affect acquisition depending
on the nature of the linguistic feature involved (consider the distinction
between resilient and fragile features, for example). We are still a long way
from explaining how input interacts with the learner’s internal cognitive
288 External factors

mechanisms to shape the course of language acquisition, and even further

from being able to assign any weighting to external as opposed to internal
factors. In all likelihood, input combines with other factors such as the
learner’s LI, the learner’s communicative need to express certain meanings
and the learner’s internal processing mechanisms. No explanation of L2
acquisition will be complete unless it includes an account of the role of input,
but, as Sharwood Smith (1985: 402) has noted input should be seen as just
one in a ‘conspiracy of factors’.

Notes

1 The distinction between ‘input text’ and ‘input discourse’ has been bor¬
rowed from Widdowson’s (1974) more general distinction between ‘text’
and ‘discourse’.
2 In Ellis 1988a, I also made the point that much of native-speaker usage is
variable (for example, English copula has a full and contracted form) and
the learner’s target, therefore, is to achieve the same pattern of variability
as that found in native-speaker usage. However, much of SLA research
has been based on the assumption that the learner’s target consists solely
of categorical rules (such as the 3rd person -s rule in English) that specify
the conditions requiring the obligatory use of a single linguistic form.
3 Snow (1986) makes the point that in order to consider to what extent
there is ‘fine-tuning’ it is essential to examine correlations between child
variables and complexity of caretaker talk at the right stage of develop¬
ment. She also discusses a number of other factors that may attenuate the
relationship (for example, fine-tuning may be situation-specific).
4 The claim that children receive no corrective feedback directed at their
grammatical errors is based on only one study, Brown and Hanlon
(1970). As Snow (1986) notes, this study used a very narrow definition of
corrective feedback (i.e. explicit corrections). Children may be able to use
other clues in the caretaker’s speech, such as caretaker responses that in¬
dicate the need for negotiation.
5 It should be noted, however, that in some cultures adult caretakers do
seem to give something resembling ‘language lessons’. Ochs and Schieffe-
lin (1984) give examples of how Kaluli mothers in Papua-New Guinea re¬
quire their children to engage in imitation exercises without any attention
to the meaning of utterances being imitated.
6 The term ‘foreigner talk’ was initially used by Eerguson (1971) to refer
solely to the kind of ungrammatical talk that native speakers sometimes
address to non-native speakers. Arthur et al. (1980) use the term ‘for¬
eigner register’ to refer to grammatical FT while the term ‘foreigner talk
discourse’ has been used to refer to interactionally modified FT. Increas¬
ingly, though, ‘foreigner talk’ is used as a general cover term for the
 Input and interaction and second language acquisition 289

modified language native speakers use with non-native speakers, and this
is how the term is used in this chapter.
7 The distinction between discourse management and discourse repair of
communication breakdowns mirrors Long’s (1983a) distinction between
‘strategies’ for avoiding trouble and ‘tactics’ for dealing with trouble
when it occurs.
8 This problem is acknowledged by Larsen-Freeman and Long (1991), who
point out that the discourse moves that perform functions such as clari¬
fication requests often have ‘multiple functions and also multiple realiza¬
tions’ (1991: 129). The need for ‘finer grained analyses’, therefore, is
recognized by at least one researcher (Long) who has been actively in¬
volved in researching the negotiation of meaning.
9 Goldin-Meadow (1982) distinguished resilient and fragile properties in
the sign language of deaf children. The former are those properties such as
recursion and word order that appear to develop irrespective of environ¬
mental conditions, whereas the latter (for example, plural and verb tense)
need special care.
10 Another purpose of Gass and Lakshmanan’s study was to show the dan¬
gers of conducting studies based on descriptions of language which ass¬
ume that the input to the learner is target-like. They comment:
‘considering principles of UG, or any other principles, devoid of context is
insufficient and often misleading in accounting for how L2 grammars
develop’.
11 Although Krashen claims that speaking does not contribute directly to ac¬
quisition, he does allow for an indirect contribution. He writes:

... output aids acquisition indirectly by encouraging Cl (comprehens¬

ible input), via conversation. When you speak it invites others to talk to
you. Moreover, as you speak your output provides your conversational
partner with information about your competence and whether he or
she is communicating successfully. This information helps your con¬
versational partner adjust the input to make it more comprehensible.
(Krashen 1989: 456)

12 The premodified input was based on negotiated input and was designed
to incorporate the main features of it.
13 Although both Long and Krashen claim that comprehensible input is ne¬
cessary for acquisition, both acknowledge that it is not sufficient. Learn¬
ers not only need access to input they can understand; they also need to be
affectively disposed to ‘let it in’ (for example, to be motivated to learn).
14 The main difference in the results obtained by Tanaka and Yamazaki is
that whereas Tanaka’s subjects in the interactionally modified input
group maintained their gains in vocabulary in the follow-up test, Yama-
zaki’s subjects did not. The explanation for this difference may lie in the
attitudes of the subjects. Tanaka’s subjects were not highly motivated to
290 External factors

learn English, whereas Yamazaki’s were—largely because they needed to

do well in English to gain entry to the universities of their choice. Yama-
zaki was able to show that many of her subjects took efforts to learn the
new vocabulary in their own time between the post-test and the follow-up
test.
15 Formal correction does, of course, occur in classroom language learning.
Chapter 14 reviews the research that has investigated the effect it has on
learning.
16 This kind of ‘focused meaning negotiation’ can be distinguished from a
more traditional grammar exercise by the fact that the learners treat the
activity as one calling for ‘use’ rather than ‘usage’; that is, they concen¬
trate on conveying a message and not on performing a particular gram¬
matical feature accurately.
17 Nobuyoshi and I noted that the one experimental learner who failed to
improve past tense accuracy was adept at making himself understood in
other ways. He manifested what Meisel, Clahsen, and Pienemann (1981)
refer to as a ‘communicative orientation’.

Further Reading
There is a rich literature dealing with input/interaction and L2 acquisition. A
good starting point is to read one or two surveys of this literature:
R. Young, ‘Input and interaction.’ Annual Review of Applied Linguistics
(1989) 9:122-34.
D. Larsen-Freeman, and M. Long, An Introduction to Second Language Ac¬
quisition Research, Chapter 5 (Longman, 1991).
The role of input and interaction in LI acquisition is considered in:
C. Snow and C. Ferguson (eds.), Talking to Children: Language Input and
Acquisition (Cambridge University Press, 1977).
G. Wells, Language Development in the Pre-school Years (Cambridge Uni¬
versity Press, 1985).
C. Srfow, ‘Conversations with children’ in P. Fletcher and M. Garman (eds.),
Language Acquisition, 2nd edition (Cambridge University Press, 1986).
For descriptive studies of FT see:
M. Clyne (ed.), Special issue on ‘Foreigner Talk’ of International Journal of
Sociology of language (1981) 28.
D. Larsen-Freeman (ed.), Discourse Analysis in Second Language Research
(see articles by Vander Brook, Schlue and Campbell, Arthur et ah, and Gas-
kill) (Newbury House, 1980).
S. Gass and C. Madden (eds.), Input in Second Language Acquisition (see art¬
icles by Kleifgen, Ellis, Wesche and Ready, Hirvonen, Gass and Varonis,
Long, and Lightbown and d’Anglejan) (Newbury House, 1985).
M. Long, ‘Native-speaker/non native-speaker conversation and the negoti¬
ation of meaning.’ Applied Linguistics (1983) 4: 126-41.
 Input and interaction and second language acquisition 291

For the effects of input/interaction on comprehension and acquisition see:

S. Krashen, The Input Hypothesis (Longman, 1985).
K. Parker and C. Chaudron, ‘The effects of linguistic simplifications and
elaborative modifications on L2 comprehension.’ University of Hawaii
Working Papers in ESL (1987) 6: 107-33.
M. Long, ‘Native-speaker/non-native-speaker conversation in the second lan¬
guage classroom’ in M. Clarke and J. Handscombe (eds.), On TESOL ‘82:
Pacific Perspectives on Language Learning and Teaching (TESOL, 1983).
T. Pica, ‘The textual outcomes of native speaker-non-native speaker negotia¬
tion: What do they reveal about second language learning’ in C. Kramsch and
S. McConnell-Ginet (eds.), Text and Context: Cross-Disciplinary Perspect¬
ives on Language Study (D. C. Heath and Company, 1992).
L. White, ‘Against comprehensible input: the Input Hypothesis and the deve¬
lopment of second-language competence.’ Applied Linguistics (1987) 8:
95-110.
 '

.
PART FOUR

Explaining second language acquisition:

internal factors
Introduction

Whereas the chapters in the last section endeavoured to explain how learners
learn a second language, why learners vary in the speed with which they learn,
and why most learners fail to achieve full target-language competence in
terms of external factors, the chapters in Part Three will examine how an ac¬
count of learner-internal mechanisms can increase our understanding of these
issues. We will attempt to go inside the ‘black box’ of the learner’s mind.
Basic to our study of learner-internal mechanisms is the distinction be¬
tween cognitive and mentalist explanations of L2 acquisition.1 This distinc¬
tion concerns the nature of the internal mechanisms responsible for
interlanguage development. A ‘cognitive’ view considers the mechanisms to
be general in nature. It sees the process of learning a language—whether a
first or a second one—as essentially the same as any other kind of learning.
Language learning engages the same cognitive systems—perception, mem¬
ory, problem solving, information processing, etc.—as learning other types
of knowledge. In a cognitive view, language learning is treated as ‘skill learn¬
ing’, analogous to learning how to ride a bicycle or play a violin, although
probably more complex.
Key terms in a cognitive account of L2 acquisition are ‘process’ and ‘strat¬
egy’. These are sometimes used interchangeably in the literature, but it is use¬
ful to distinguish between them. The term ‘process’ can be found in such
phrases as the ‘learning process’, the ‘comprehension process’, and the ‘pro¬
duction process’. Following Klaus and Buhr (1976: 990), cited in Frerch and
Kasper (1983b), we will use ‘process’ in this sense to refer ‘the dynamic se¬
quence of different stages of an object or system’. Implicit in this definition is
that (1) L2 learning and L2 use both involve a series of steps which are system¬
atic in that they occur whenever learners try to learn or use the L2, and (2)
these steps are, to some extent at least, common to all learners and users. The
term ‘strategy’ is more difficult to tie down, as it is used with somewhat differ¬
ent meanings. In general, the term is used to refer to some form of activity,
mental or behavioural, that may occur at a specific stage in the overall process
of learning and communicating. In Fserch and Kasper’s framework, strategies
are ‘plans’ for accomplishing some part of the process. Other aspects of stra¬
tegies are more controversial, however. Bialystok (1978) treats strategies as
‘optional mental activities’, in contrast to processes which are ‘obligatory’,
but this distinction has not always been followed. Fasrch and Kasper consider
‘problem orientedness’ and ‘consciousness’ to be the two defining criteria of
296 Internal factors

‘communication strategies’, but others treat strategies as not necessarily re¬

lated to problems (in learning or communication) and, at least sometimes, as
subconscious.
A mentalist account of L2 acquisition is based on the distinction between
‘competence’ and ‘performance’, a distinction that we have already con¬
sidered in various parts of this book (see Chapter 1, in particular). Bialystok
and Sharwood Smith (1985) present a closely argued application of this dis¬
tinction to L2 acquisition in their ‘knowledge’ and ‘control’ model (see the
diagram below). Knowledge refers to ‘the way in which the language system
is represented in the mind of the learner’ and control to ‘the processing system
for controlling knowledge during actual performance’ (1985: 104). The
knowledge dimension is further divided into linguistic competence and prag¬
matic competence (knowledge of how to use linguistic knowledge for com¬
municating). The disparity between native-speaker and learner proficiency
can reflect (1) differences in knowledge (linguistic or pragmatic), (2) differ¬
ences in control, or (3) both. Similarly, L2 development can consist of
changes in the learner’s knowledge systems, an increase in control of this
knowledge system, or both. This model allows for the possibility of studying
learners’ grammatical systems (i.e. ‘knowledge’) independently of their use in
communication (i.e. ‘control’).

Linguistic knowledge
Output
Pragmatic

Grammatical Control procedures

___^ production
competence reception

competence

The relationship between control procedures and two aspects of linguistic

knowledge (Bialystok and Sharwood Smith 1985: 106)

The main difference between cognitive and mentalist accounts of language

learning lies in how they view the relationship between ‘knowledge’ and ‘con¬
trol’. In a cognitive account, knowledge is not seen as distinct from control.
There is no clear dividing line between what learners know and what they can
do with their knowledge, as the way linguistic knowledge is represented in the
mind of the learner is influenced by the uses to which this knowledge is put. In
contrast, a mentalist account of L2 acquisition treats ‘knowledge’ and ‘con¬
trol’ as separate, and also sees the mechanisms responsible for the develop¬
ment of each as distinct. The central claim of a mentalist theory of language
acquisition is that linguistic knowledge constitutes a separate cognitive
 Internal factors 297

faculty, independent of the other cognitive systems involved in the use of this
knowledge.
The study of LI transfer is the starting point for this section, mainly be¬
cause, historically speaking, it constituted the first attempt to provide an ex¬
planation for L2 acquisition. Initially, transfer was viewed in behaviourist
terms, but increasingly it has been treated as a cognitive strategy. The follow¬
ing chapter examines a number of cognitive accounts of L2 acquisition. These
are varied in nature but they have in common the assumption that L2 learning
constitutes a ‘process’, governed by mechanisms of a general cognitive
nature. In the final chapter in this section we will adopt a mentalist stance
by examining how theories of language have contributed to the study of
L2 acquisition.

Notes
1 This distinction is similar but not identical to Bialystok’s (1990b) distinc¬
tion between ‘processing’ and ‘competence’ theories. Bialystok defines the
former as ‘descriptions of methods of storage and means of accessing rules’
and the latter as ‘descriptions of knowledge of rules for linguistic structure’
(1990b: 637). In fact, however, both cognitive and mentalist explanations
of L2 acquisition make reference to ‘rules’. The difference lies in the nature
of the mechanisms that are proposed to describe the rules.
'
8 Language transfer

Introduction

This chapter begins the study of the cognitive structures involved in second
language acquisition by considering how the learner’s existing linguistic
knowledge influences the course of L2 development. This constitutes a suit¬
able starting point, not so much because the learner’s existing linguistic
knowledge is the major factor, but because, historically speaking, it was the
first factor to receive serious attention.
Behaviourist views of language learning and of language teaching were pre¬
dominant in the two decades following the second world war. These views
drew on general theories of learning propounded by psychologists such as
Watson (1924), Thorndike (1932), and Skinner (1957). Dakin (1973) ident¬
ifies three general principles of language learning derived from these theories.
According to the law of exercise, language learning is promoted when the
learner makes active and repeated responses to stimuli. The law of effect em¬
phasizes the importance of reinforcing the learners’ responses by rewarding
target-like responses and correcting non-target-like ones. The principle of
shaping claims that learning will proceed most smoothly and rapidly if com¬
plex behaviours are broken down into their component parts and learnt bit
by bit. Underlying these principles was the assumption that language learn¬
ing, like any other kind of learning, took the form of habit formation, a ‘habit’
consisting of an automatic response elicited by a given stimulus. As Brooks
(1960) put it:

The single paramount fact about language learning is that it concerns, not
problem solving, but the formation and performance of habits (1960: 49).

Learning was seen to take place inductively through ‘analogy’ rather than
‘analysis’.
According to behaviourist theories, the main impediment to learning was
interference from prior knowledge. Proactive inhibition occurred when old
habits got in the way of attempts to learn new ones. In such cases, the old
habits had to be ‘unlearnt’ so that they could be replaced by new ones. In the
case of L2 learning, however, the notion of ‘unlearning’ made little sense, as
learners clearly did not need to forget their LI in order to acquire an L2, al¬
though in some cases loss of the native language might take place eventually.
For this reason, behaviourist theories of L2 learning emphasized the idea of
300 Internal factors

‘difficulty’, defined as the amount of effort required to learn an L2 pattern.

The degree of difficulty was believed to depend primarily in the extent to
which the target language pattern was similar to or different from a native-
language pattern. Where the two were identical, learning could take place
easily through positive transfer of the native-language pattern, but where
they were different, learning difficulty arose and errors resulting from negat¬
ive transfer were likely to occur. Such errors or ‘bad habits’ were considered
damaging to successful language learning because they prevented the forma¬
tion of the correct target-language habits. The commonly held view was that
‘like sin, error is to be avoided and its influence overcome, but its presence to
be expected’ (Brooks 1960).
These behaviourist views have now been discredited. Chomsky’s (1959)
review of Skinner’s Verbal Behaviour set in motion a re-evaluation of many
of the central claims. The dangers of extrapolating from laboratory studies of
animal behaviour to the language behaviour of humans were pointed out.
The terms ‘stimulus’ and ‘response’ were exposed as vacuous where language
behaviour was concerned. ‘Analogy’ could not account for the language
user’s ability to generate totally novel utterances. Furthermore, studies of
children acquiring their LI showed that parents rarely corrected their chil¬
dren’s linguistic errors, thus casting doubt on the importance of ‘reinforce¬
ment’ in language learning (see Chapter 7). These studies suggested that
language acquisition was developmental in nature, driven as much, if not
more, from the inside as from the outside. As we shall see, the demise of beha¬
viourist accounts of language learning led to a reconsideration of the role of
the LI in L2 learning.
This chapter is primarily concerned with the cognitive accounts of lan¬
guage transfer that have now superseded the behaviourist views outlined
above. It begins with a discussion of terminological issues, the term ‘transfer’
itself being somewhat controversial, and then goes on to describe a number of
different ways in which transfer manifests itself. The chapter then examines
the Contrastive Analysis Hypothesis, the problems identified with it, and its
current status in SLA research. There follows a brief consideration of the min¬
imalist position regarding the role of the LI, with particular reference to the
views of Dulay, Burt, and Krashen. The main section of the chapter examines
various constraints on transfer in order to account for the fact that transfer is
sometimes apparent and sometimes not—the ‘now you see it, now you don’t’
nature of language transfer (Kellerman 1983). This provides the basis for a
general framework to account for the role of prior linguistic knowledge. The
chapter concludes with a discussion of some of the problems inherent in the
study of transfer.
In subsequent chapters we will need to return to the issue of LI transfer, as
no theory of L2 acquisition that ignores the learner’s prior linguistic know¬
ledge can be considered complete. In particular, the section dealing with the
 Language transfer 301

Competition Model in Chapter 9 and that dealing with UG-based empirical

studies of L2 acquisition in Chapter 10 will further address the role of the LI.

Terminological issues

As we have seen, the terms ‘interference’ and ‘transfer’ are closely associated
with behaviourist theories of L2 learning. However, it is now widely accepted
that the influence of the learner’s native language cannot be adequately
accounted for in terms of habit formation. Nor is transfer simply a matter of
interference or of falling back on the native language. Nor is it just a question
of the influence of the learner’s native language, as other previously acquired
‘second’ languages can also have an effect. This suggests that the term ‘LI
transfer’ itself is inadequate. Sharwood Smith and Kellerman (1986) have
argued that a superordinate term that is theory-neutral is needed and suggest
crosslinguistic influence. They comment:

... the term ‘crosslinguistic influence’... is theory-neutral, allowing one to

subsume under one heading such phenomena as ‘transfer’, ‘interference’,
‘avoidance’, ‘borrowing’ and L2-related aspects of language loss and thus
permitting discussion of the similarities and differences between these
phenomena (1986: 1).

There is much to be said for this proposal, not least the precise definition of
‘transfer’ that it permits. Kellerman (1987) has suggested that the term be
restricted to ‘those processes that lead to the incorporation of elements from
one language into another’ (1987: 3). However, as is often the case in ter¬
minological disputes, usage does not always conform to reason, and the term
‘transfer’ has persisted, although its definition has now been considerably
broadened to include most of the crosslinguistic phenomena that Kellerman
and Sharwood Smith consider to be in need of attention. Odlin (1989) offers
this ‘working definition’ of transfer as a basis! for his own thoughtful treat¬
ment of such phenomena:

Transfer is the influence resulting from the similarities and differences be¬
tween the target language and any other language that has been previously
(and perhaps imperfectly) acquired (1989: 27).

Transfer viewed in this way is, of course, far removed from the original use of
the term in behaviourist theories of language learning. This definition,
although somewhat vague (as Odlin admits), provides an adequate basis for
the material to be considered in this chapter.

The manifestations of transfer

In traditional accounts of language transfer, the research focus was placed on
the errors that learners produce. Errors occurred as a result of the negative
302 Internal factors

transfer of mother tongue patterns into the learner’s L2. It is possible to

identify a number of other manifestations of transfer, however, three of
which will be considered here: facilitation, avoidance (or underproduction),
and over-use.

Errors (negative transfer)

Learners’ errors were considered in detail in Chapter 2, where the procedure

for conducting an error analysis was described. We noted that a substantial
amount of empirical work in SLA research has been devoted to establishing to
what extent errors are the result of transfer (i.e. interference) or are intra-
lingual in nature (i.e. the result of general processes of language development
similar to those observed in LI acquisition). Table 8.1, from Ellis 1985a, il¬
lustrates the considerable variance in the proportion of transfer errors re¬
ported by different investigators. Whereas Dulay and Burt (1973) report that
transfer accounted for only 3 per cent of the errors in their corpus of Spanish¬
speaking learners’ L2 English, Tran-Chi-Chau (1975) reports 51 per cent in
adult, Chinese-speaking learners’ English. As we saw in Chapter 3, one of the
main reasons for this variation is the difficulty in determining whether an
error is the result of transfer or intralingual processes.

Study % of interference errors Type of learner

Grauberg 1971 36 First language German—

adult, advanced
George 1972 33 (approx) Mixed first languages—
adult, graduate
Dulay and Burt 1973 3 First language Spanish—
children, mixed level
Tran-Chi-Chau 1975 51 First language Chinese—
adult, mixed level
Mukkatesh 1977 23 First language Arabic—
adult
Flick 1980 31 First language Spanish—
adult, mixed level
Lott 1983 50 (approx) First language Italian—
adult, university

Table 8.1: Percentage of interference errors reported by various studies of L2

English grammar (from Ellis 1985a: 29)

Facilitation (positive transfer)

The learner’s LI can also facilitate L2 learning. Odlin (1989) points out that
the facilitative effects can only be observed when learners with different
 Language transfer 303

native languages are studied and learner comparisons are carried out. Facil¬
itation is evident not so much in the total absence of certain errors—as would
be expected on the basis of behaviourist notions of positive transfer—but
rather in a reduced number of errors and, also, in the rate of learning.
The facilitative effect of the LI can also be adduced by certain types of
U-shaped behaviour1 (see Kellerman 1985a). Learners may sometimes pass
through an early stage of development where they manifest correct use of a
target-language feature if this feature corresponds to an LI feature and then,
subsequently, replace it with a developmental L2 feature before finally re¬
turning to the correct target-language feature. In such a case, the facilitative
effect is evident in the early stages of acquisition, before the learner is ‘ready’
to construct a developmental rule. The ‘re-learning’ of the correct target-
language rule occurs when learners abandon the developmental rule as they
come to notice that it is incompatible with the input. Evidence for such a facil¬
itative effect, of course, can best be obtained through the longitudinal study
of individual learners.
Two studies of relative clauses illustrate how transfer can have a facilitative
effect. Gass (1979; 1983) investigated 17 adult learners of L2 English with di¬
verse language backgrounds. Data relating to a number of structural aspects
of relative clauses were collected, but only one—pronoun retention—
provided clear evidence of transfer effects. Gass divided her subjects into two
groups according to whether their LI allowed pronoun retention (for
example, Persian and Arabic) or did not allow it (for example, French and
Italian). She found that learners in the first group were much more likely to
accept sentences like:

* The woman that I gave a book to her is my sister.

as grammatical than learners in the second group. Those learners in the se¬
cond group, whose languages resembled English in not permitting pronoun
retention, also made fewer errors in a sentence-joining task. Interestingly,
though, facilitative transfer effects were not evident in all relative clause
types, there being no difference between the two groups in the case of clauses
where the pronoun functioned as subject, genitive, or object of comparison in
the sentence-joining task. This led Gass to claim that transfer interacts with
other, universal factors—an important point which we will take up later.
The second study (Hyltenstam 1984) investigated relative clauses in L2
Swedish, a language that does not permit pronoun retention. The subjects
were 45 adult learners from five language groups. These languages differed in
the extent to which they manifested pronominal copies in relative clauses.
Hyltenstam used pictures to elicit oral sentences for each relativizable func¬
tion. The results showed that pronominal copies occurred in the speech pro¬
duced by all the language groups irrespective of whether their LI permitted
retention or not, but that the frequency of the copies varied according to lan¬
guage group. The Persian learners produced the most copies, followed by the
304 Internal factors

Greek, with the Spanish and Finnish learners producing the fewest. This
order corresponded exactly to that predicted on the basis of the structural
properties of the learners’ native languages.
The facilitative effect of the LI is evident in other aspects of L2 acquisition.
In many cases, this is obvious, as when two languages share a large number of
cognates (for example, English and French), thus giving the learners a head
start in vocabulary. Chinese learners of L2 Japanese have an enormous
advantage over English learners because of the similarity of the Chinese and
Japanese writing systems. They are able to make use of written as well as spo¬
ken input straight away. The positive contribution of the LI has been ignored
by some researchers who, in staking out a minimalist transfer position, have
chosen to attend only to learner errors.

Avoidance

Learners also avoid using linguistic structures which they find difficult be¬
cause of differences between their native language and the target language. In
such cases, the effects of the LI are evident not in what learners do (errors) but
in what they do not do (omissions). The classic study of avoidance, which we
considered briefly in Chapter 3, is Schachter (1974). Schachter found that
Chinese and Japanese learners of L2 English made fewer errors in the use of
relative clauses than Persian or Arabic learners because they produced far
fewer clauses overall (see Table 8.2). In Levenston’s (1971) terms, ‘under¬
representation’ occurs, which, as Levenston (1979) points out, is best charac¬
terized as an aspect of ‘use’ rather than ‘acquisition’.

Correct Error Total Percentage of Errors

Persian 131 43 174 25

Arab 123 31 154 20
Chinese 67 9 76 12
Japanese 58 5 63 8
American 173 0 173 0

Table 8.2: Relative clause production in five languages (Schachter 1974: 209)

The difficulty for the Chinese and Japanese learners may lie in the fact that
while their mother tongues are left-branching (i.e. nouns are pre-modified):

Aki ga kinoo honya de kat-ta hon ...

(Aki NOM yesterday bookstore LOC buy-PAST) book ...2

English is primarily right-branching (i.e. nouns are post-modified):

The book which Aki bought yesterday at the bookstore ....

 Language transfer 305

Persian and Arabic are also right-branching. Schachter hypthesizes that it is

this factor which leads Chinese and Japanese learners of L2 English to avoid
using relative clauses.
The identification of avoidance is not an easy task. Seliger (1989) points
out that it is only possible to claim that avoidance has taken place if the
learner has demonstrated knowledge of the form in question, and if there is
evidence available that native speakers of the L2 would use the form in the
context under consideration. In other words, it only makes sense to talk of
avoidance if the learners know what they are avoiding. Kamimoto, Shimura,
and Kellerman (1992) argue that even demonstrating knowledge of a struc¬
ture is not sufficient. Hebrew speakers of English may know how to use the
passive, but their infrequent use of it in the L2 may simply reflect the general
preference for active over passive in LI Hebrew rather than avoidance (i.e.
negative transfer). Thus, it is also necessary to demonstrate that the structure
is not under-used simply because the equivalent structure is rare in the LI.
Avoidance is a complex phenomenon. Kellerman (1992) attempts to sort
out the complexity by distinguishing three types. Avoidance (1) occurs when
learners know or anticipate that there is a problem and have at least some,
sketchy idea of what the target form is like. This is the minimum condition for
avoidance. Avoidance (2) arises when learners know what the target is but
find it too difficult to use in the particular circumstances (for example, in the
context of free-flowing conversation). Avoidance (3) is evident when learners
know what to say and how to say it but are unwilling to actually say it because
it will result in them flouting their own norms of behaviour. In all three cases
it is clear that much more is involved than the learner’s LI. The extent of
learners’ knowledge of the L2 and the attitudes learners hold toward their
own and the target-language cultures act as factors that interact with LI
knowledge to determine avoidance behaviour.

Over-use

The over-use or ‘over-indulgence’ (Levenston 1971) of certain grammatical

forms in L2 acquisition can occur as a result of intralingual processes such as
overgeneralization (see Chapter 2). Lor example, L2 learners have often been
observed to overgeneralize the regular past tense inflection to irregular verbs
in L2 English (for example, ‘costed’). Similarly, learners may demonstrate a
preference for words which can be generalized to a large number of contexts
(Levenston 1979). Over-use can also result from transfer—often as a con¬
sequence of the avoidance or underproduction of some ‘difficult’ structure.
Japanese learners of English, for example, may overproduce simple sentences
and may even be encouraged to do so, as this professional advice from a
Japanese translator shows:
306 Internal factors

Translate a main clause with a relative clause in English into two main
clauses and connect them with conjunctions. (Kamimoto, Shimura, and
Kellerman 1992: 268).

Over-use as a result of transfer is also evident at the discourse level.

Olshtain (1983) examined the apologies offered by 63 American college stu¬
dents learning L2 Hebrew in Israel. She was able to show that the native
speakers of English used more direct expressions of apology than native
speakers of Hebrew and that they tended to transfer this into L2 Hebrew.
This tendency was evident across situations (see Chapter 5 for a fuller discus¬
sion of the research on apologies).
Over-use of linguistic and discourse features as a result of LI influence is
probably more common than generally acknowledged. As in the case of
avoidance, it can only be detected by comparing groups of learners with dif¬
ferent Lis.

Summary and final comment

Current discussions of transfer emphasize the necessity of considering the

multiple ways in which LI influence can exert itself. It is clearly insufficient to
focus exclusively on production errors, as many of the subtle manifestations
of transfer will be missed. Transfer scholars, therefore, also look for evidence
of facilitation (positive transfer), avoidance, and over-use. They recognize,
however, that all these aspects of L2 use are likely to demonstrate a variety of
factors at work, only one of which is transfer. Thus, they emphasize the im¬
portance of looking for how LI knowledge interacts with other factors.

Difference and difficulty: the role of contrastive analysis

Although behaviourist views of transfer are largely defunct, the central no¬
tions that they gave rise to—‘difference’ and ‘difficulty’—are still very much
alive. We will begin this section with a consideration of early formulations of
the Contrastive Analysis Hypothesis, consider some of the problems it gave
rise to and then offer a reconsideration of the ‘difference = difficulty’ equation
and of the role of contrastive analysis.
The Contrastive Analysis Hypothesis (CAH) as formulated by Lado (1957)
was based on this assumption:

... the student who comes into contact with a foreign language will find
some features of it quite easy and others extremely difficult. Those elements
that are similar to his native language will be simple for him, and those ele¬
ments that are different will be difficult. (1957: 2)

Lado began his book, Linguistics Across Cultures, by referring to the work of
Haugen (1956) and Weinreich (1953). This provided evidence that when lan¬
guages come into contact in bilingual situations, many of the ‘distortions’
 Language transfer 307

that arise are traceable to differences in the languages involved. Lado’s book
not only laid out the theoretical bases of the CAH but also described the tech¬
nical procedures needed to carry out the detailed contrastive analyses that
were considered necessary for the preparation of ‘scientific’ teaching
materials.
The 1960s saw a spate of contrastive analyses involving the major Euro¬
pean languages (for example, Stockwell and Bowen 1965; Stockwell, Bowen,
and Martin 1965). They were designed to provide course developers with
information regarding where the differences and hence the difficulties lay
for learners of different language backgrounds. The analyses were based on
surface ‘structuralist’ descriptions of the two languages concerned. The pro¬
cedure followed involved (1) description (i.e. a formal description of the
two languages was made), (2) selection (i.e. certain areas or items of the two
languages were chosen for detailed comparison), (3) comparison (i.e. the
identification of areas of difference and similarity), and (4) prediction (i.e.
determining which areas were likely to cause errors).
The scholars of the 1960s recognized different kinds of ‘difference’ and
also attributed to them different degrees of ‘difficulty’. Table 8.3 below pro¬
vides a simplified version of the hierarchy of difficulty produced by Stockwell,
Bowen, and Martin (1965). It shows that ‘difficulty’ will be greatest when
there is a split and least in the case of coalesced forms. No difficulty arises
when there is a complete correspondence of items in the two languages.

Type of difficulty LI: English Example

L2: Spanish

1 Split .x
x
‘for’ is either por or para

2 New o o grammatical gender

3 Absent X o ‘do’ as a tense carrier
4 Coalesced X

X
‘his/her’ is realized as a single form su

y
5 Correspondence X X -ing, -ndo as complement with verbs of
perception, e.g. ‘1 saw the men running’;
vi a los hombres corriendo.

Table 8.3: Simplified version of the hierarchy of difficulty (based on information

given in Stockwell, Bowen, and Martin 1965)

In its strongest form, the CAH claimed that all L2 errors could be predicted by
identifying the differences between the learners’ native language and the tar¬
get language. Lee (1968: 180), for instance, stated that ‘the prime cause, or
308 Internal factors

even the sole cause, of difficulty and error in foreign language learning is in¬
terference coming from the learner’s native language’. However, given the
empirical evidence that soon became available suggesting that many errors
were not the result of transfer (see Table 8.1), a weaker form of the hypothesis
was proposed (Wardhaugh 1970). According to this, only some errors were
traceable to transfer, and contrastive analysis could be used only a posteriori
to explain rather than predict. In other words, contrastive analysis needed to
be used hand in hand with error analysis.
The CAH had its heyday in the 1960s, but gradually fell out of favour in the
1970s. There were several reasons for this. The strong form became unten¬
able when it was shown that many errors were not caused by transfer (see
Dulay and Burt 1974a) and that many errors predicted by contrastive ana¬
lysis did not actually occur (see Jackson and Whitnam 1971). However, the
weak form is also problematic. First, an a posteriori contrastive analysis is
something of a ‘pseudo procedure’ (James 1980) in the sense that it makes
little sense to undertake a lengthy comparison of two languages simply to
confirm that errors suspected of being caused by transfer are indeed so. James
argued that a contrastive analysis was only worthwhile if it was predictive.
Schachter (1974) also argued that an a priori contrastive analysis was needed
in order to identify those areas of the L2 system which learners might try to
avoid. Neither version of the CAH was convincing, therefore; the strong ver¬
sion was theoretically untenable and the weak version was impractical and
inadequate. It was not surprising to see contrastive analysis lose ground to
error analysis in the 1970s.
Despite these problems, transfer continued to be of interest to SLA re¬
searchers. To explain how researchers accommodated to the growing evid¬
ence that transfer was, at best, only a partial explanation of learning
difficulty, we need to reconsider the two basic claims of the CAH:

(1) The level of difficulty experienced by the learner will be directly related
to the degree of linguistic difference between the LI and L2.
(2) Difficulty will manifest itself in errors; the greater the difficulty, the
' more frequent the errors.

Kellerman (1987) has argued that the only viable definition of a ‘difficult’
structure is one that learners fail to learn despite plentiful evidence for the ex¬
istence of the structure in the input. There are many possible explanations for
a structure being ‘difficult’—for example, its saliency, its communicative
value to the learner, the extent to which it is marked or unmarked, or the ease
with which it can be processed in production or comprehension (see Chapter
9). Included in these possible explanations is the extent to which the structure
is similar to or different from a comparable structure in the learner’s LI or an¬
other previously learnt L2. The problem with the CAH was its failure to ac¬
knowledge sources of difficulty other than the learner’s LI. While it has been
shown to be clearly incorrect to claim that L1/L2 differences will lead to
 Language transfer 309

difficulty, it has also become clear that they might do. The trick is to show
when they do and when they do not. As Kellerman (1987) puts it, there is a
need to establish the ‘potential for transfer’ of a given structure and to identify
the nature of the constraints that govern its transferability. The second claim
of the CAH is, in fact, not a necessary one—‘errors’ are but one aspect of the
learning process and, as we saw in the previous section, difficulty can mani¬
fest itself in other forms.
In short, the problem with the CAH is that it is too simplistic and too re¬
strictive. The solution, as many researchers have come to recognize, lies not in
its abandonment but in careful revision and extension. These researchers con¬
tinue to make use of contrastive analysis, but only as a tool for identifying po¬
tential areas of difficulty (Fisiak 1981a). It was generally recognized that the
claims of a contrastive analysis needed to be subjected to careful empirical in¬
quiry. As Selinker (1969) pointed out, ‘experimentation in language research’
is needed ‘whenever it becomes clear that extrapolation from pure linguistic
research is improper’. Similarly, Nemser (1971) saw the ‘direct and system¬
atic examination of learner speech’ as a ‘prerequisite for the validation of
both the strong and weak claims of the contrastive approach’. Thus, the diffi¬
culties with the CAH set the scene for the detailed investigation of learner
language, as in Selinker’s study of Hebrew learners of English. Subsequently,
researchers undertook the kind of learner comparisons which, as we have
seen, are necessary to provide unambiguous evidence of transfer in all its
manifestations.

The minimalist position

The empirical research in the late 1960s and 1970s, which we considered in
Chapters 2 and 3, led to two rather different ways of accounting for the role
of the LI. In one, transfer was treated as one of several processes involved in
L2 acquisition. This was the approach of Selinker (1969; 1972), Nemser
(1971), and James (1971). Selinker (1972) saw ‘language transfer’ as one of
the five processes responsible for fossilization, placing it, in fact, at the top of
his list. Nemser (1971) claimed that learner speech was ‘structurally organ¬
ized’ in the sense that it constituted a system in its own right. However, like
Selinker, he saw the learner’s native language as one of the major determin¬
ants of these ‘approximative systems’. James (1971) pointed out that con¬
trastive analysts had never claimed that LI interference was the sole source of
error. He also noted, prophetically, that ‘it will probably turn out that many
of the errors which are now not traceable to the LI, and are therefore attrib¬
uted to L2 overgeneralization, will... be recognized as errors of interference’
(1971: 56). As we will see later, it was such a view of transfer that came to
dominate and to lead researchers to investigate how transfer interacted with
other, non-Ll based strategies. In this section, however, we will focus on the
second way of dealing with transfer—the minimalist approach.
310 Internal factors

The minimalist position sought to play down the importance of the LI and
to emphasize the contribution of universal processes of language learning,
such as hypothesis-testing. It found expression in three ways: the dismissal of
evidence of the pervasiveness of interference which came from studies of lan¬
guage contact situations, the conduct of empirical studies designed to test the
CAH, and the development of alternative theoretical arguments.

Interference in language contact situations and second language

acquisition
Although the experience of many teachers testified to the existence of transfer
errors, there were, in fact, hardly any empirical studies of transfer in L2 learn¬
ing. Consequently, the CAH relied for supportive evidence on the work of
Weinreich (1953) and Haugen (1956) in bilingual contact situations. This
showed that many of the ‘linguistic distortions’ observed in the speech of
bilinguals corresponded to differences in the languages involved. However,
Dulay and Burt (1972) claimed that the concepts of ‘interference’ and ‘lin¬
guistic borrowing’ used by Weinreich and Haugen were distinct from that of
‘interference’ in the CAH and that scholars such as Lado were not justified in
citing Weinreich’s and Haugen’s research in its support. Interference in bilin¬
guals, they argued, was motivated by social factors, was bi-directional, and
increased with proficiency in the two languages; interference in language
learning was not socially motivated, was unidirectional, and decreased as the
learner became more proficient. It is not at all clear that the two uses of ‘inter¬
ference’ are as distinct as Dulay and Burt claim.3 Indeed, current definitions of
the term ‘transfer’ (see page 301 ) allow for both sociolinguistic and psycho-
linguistic LI effects. Although the two phenomena are not identical, they are
now seen to be related. Even the frequently made distinction between bor¬
rowing transfer (where the L2 influences the LI) and substratum transfer
(where the LI influences the L2) is not always clear-cut (see Thomason and
Kaufman’s (1988) study of language learning in Ethiopia, cited in Odlin
1989: 12). It is no longer possible to dismiss the evidence of transfer effects in
bilinguals as irrelevant to L2 acquisition.

Empirical research and the CAH

A number of studies of learner language carried out in the late 1960s and
early 1970s indicated that the influence of the LI was much less than that
claimed by the CAH. These studies were based on attempts to quantify the
number of transfer errors. For example, in the most frequently cited study,
Dulay and Burt (1974b) examined six structures in the L2 speech of Spanish
learners of English. They selected only those structures that could provide
 Language transfer 311

unambiguous evidence of transfer. For example, negative constructions such

as the following:

* I not have a bike.

were not included in the study, because although similar constructions are
present in Spanish:

Yo no tengo bicicleta.

they are also found in the LI acquisition of English, thus making it impossible
to decide whether they constitute examples of interference or developmental
errors. In contrast, the presence of a preclitic in sentences like:
* The dog it ate.

constitutes clear evidence of interference because, while the same construc¬

tion is found in Spanish:
El perro se lo comio.

it is not evident in the LI acquisition of English. The study showed that less
than 5 per cent of the total errors in the corpus were of the ‘interference’ type,
the rest being either ‘developmental’ or ‘unique’ (see page 60 for a definition
of these categories). Other studies have also claimed to show that the learners’
LI plays only an insignificant role. For example, Felix (1980b) examined
three syntactic structures in English-speaking children’s acquisition of L2
German and concluded that ‘interference does not constitute a major strat¬
egy’(1980b: 107).
There are reasons for believing, however, that studies like those of Dulay
and Burt and Felix seriously underestimate the role of the LI. First, it is not
always easy to decide whether an error is interference or developmental, and
this can result in estimations of transfer errors that are too conservative. As
we have seen, it can be argued that ‘no’ + verb constructions produced by
Spanish learners of English do not constitute a transfer error because the same
error is evident in the LI acquisition of English. It is possible, however, that
‘no’ + verb constructions reflect transfer and developmental processes
working in conjunction. The point is that by eliminating structures with a
potential for transfer (for example, negative constructions) and by in¬
vestigating constructions, such as preclitics, where transfer is theoretically
possible but unlikely—for reasons to be considered later—Dulay and Burt
have contrived to ensure that they find few interference errors. The
assumption that errors must be either the result of interference or intralingual
is unwarranted. Also, before transfer can be dismissed, it is necessary to
demonstrate through learner comparisons that the LI is indeed having no
effect.
As Kellerman (1987) has pointed out, researchers tend to reflect their the¬
oretical biases in what they interpret as transfer effects. He notes that Arabski
(1979) made the somewhat surprising assertion that the 974 article errors in
312 Internal factors

his Polish-English corpus were not transfer errors on the grounds that, be¬
cause Polish does not have articles, there is nothing to transfer. Clearly,
though, the absence of a structural feature in the LI may have as much impact
on the L2 as the presence of a different feature. Kellerman points out that if
Arabski had included the article errors in his corpus, the total percentage at¬
tributable to transfer would have exceeded 70 per cent (as opposed to the 50
per cent Arabski actually reported).
Another point made by Kellerman is that the contribution of the LI can
also be underestimated if insistence is placed on quantification in terms of
error tokens rather than error types. As the number of error tokens will de¬
pend on the number of obligatory occasions for the use of a particular struc¬
ture, it is obvious that wide fluctuations can occur. Kellerman gives the
example of article errors, which are likely to be more common than passive
errors for the simple reason that there is greater opportunity for their com¬
mission. Obviously, the true effect of the LI can only be established if re¬
searchers examine error types (i.e. the number of different structures in which
LI influence is evident). The minimalist positions of Dulay and Burt and Lelix
are based on counting error tokens and consider very few error types.

Word order studies of transfer

A good example of the kind of problems that can arise in trying to determine
the extent of LI influence is found in studies of basic word order. Languages
vary in (1) their basic word order (the majority of languages being VSO, SVO,
or SOV) and (2) the extent to which their basic word order is rigid or flexible
(for example, English has a rigid word order, while Russian is flexible).
Odlin (1989; 1990) provides a review of the research that has investigated
to what extent word order in interlanguage is affected by the LI. He refers to
the ‘universalist position’ advanced by Rutherford and Zobl, namely that
there is little negative transfer where basic word order is concerned. He cites
Rutherford (1983), for instance, as claiming that ‘Japanese learners of Eng¬
lish do not at any time produce writing in which the verb is wrongly placed
sentence finally’ (1983: 367). Given that Japanese is rigidly SOV and English
SVO, transfer might have been expected. Zobl (1986) similarly argues that
L2 learners do not make recourse to LI knowledge where basic word order is
concerned, except where the target language has more than one basic word
order, as in the case of Dutch and German, which have SVO (and other pat¬
terns) in main clauses and SOV in subordinate clauses. Rutherford and Zobl’s
universalist position is in direct opposition to a transfer position.
Minimalist accounts of transfer seek to explain away apparent instances of
LI effects on word order. Lor example, Turkish learners of German have
been observed to sometimes manifest constructions where the verb appears at
the end of a sentence, suggesting that LI transfer is taking place. However,
Clahsenand Muysken (1986) argue that what appears to be transfer is in fact
 Language transfer 313

a discourse strategy. Odlin (1989: 89), too, accepts that there is ‘detailed
evidence for the heavy reliance of some learners on topic-comment patterning
in the early stages of acquisition’. He cites these examples from Huebner’s
(1983) study of a Hmong refugee in Hawaii:

mii wok (As for me, I walked)

hos, ai reis (As for horses, I raced).

Learners of German, irrespective of their LI, have been found to produce sim¬
ilar utterances (see Klein 1986; Clahsen and Muysken 1986).
These studies, then, suggest that the learner’s LI has little influence on basic
word order and that, even when transfer is apparent, it can be better ex¬
plained in terms of a discourse strategy. They lend support to the minimalist
position. However, Odlin (1990) goes on to muster substantial and convinc¬
ing evidence of LI transfer where basic word order is concerned.
He provides evidence from 11 studies of substratum transfer (i.e. from the
LI to the L2) and borrowing transfer (i.e. from the L2 to the LI) of word
order rules. For example, Nagara (1972) reports that Japanese speakers of
Hawaiian pidgin produce sentences like:

Mi: cu: stoa gecc (me two store get = I got two stores)
hawai kam (Hawaii came = I came to Hawaii).

It is possible that such utterances reflect a topic-comment patterning. How¬

ever, Bickerton and Givon (1976) have demonstrated that Japanese and Fili¬
pino speakers of Hawaiian Pidgin English manifest a word order that differs
in accordance with their LI. Givon (1984) provides further evidence against a
discourse explanation. He showed that Filipino and Korean speakers of
Hawaiian pidgin displayed a preference for different word order patterns
which directly reflected their Lis. Odlin also argues convincingly that Lujan,
Minaya, and Dankoff’s (1984) study of Spanish spoken in Peru and Ecuador
provides evidence of influence from the LI word order of local Indian lan¬
guages, again dismissing Muysken’s (1984) claim that the observed OV pat¬
terns are the result of ‘stylistic considerations’. Further examples of
substratum transfer cited by Odlin are in Korean Bamboo English, a pidgin
used in contacts between American soldiers and Japanese and Korean civil¬
ians in the 1940s and 1950s, and Pidgin Fijian.
Odlin concludes that ‘there is no universal constraint on the transfer of
basic word order’ (1990: 107), contradicting the claim of Rutherford and
Zobl. Again, then, the minimalist position is not supported—even in the case
of a structure that might be considered to be highly amenable to a universalist
explanation.
However, Odlin (1990) does admit that there are relatively few instances of
basic word order transfer in the literature and suggests two principal reasons
why this might be so. First, he points out the relative lack of research on
beginner learners, in whom word order transfer is most likely. Second,
314 Internal factors

learners are likely to be highly conscious of word order as it involves the

arrangement of semantically important elements. He notes that learners are
very successful in identifying word order errors and, also, adept at imitating
the word order patterns of other languages. He reports that Korean-English
and Spanish-English bilinguals were much more consistent in judging word
order errors than article errors. Teachers are also likely to pay attention to
word order violations. The high level of awareness that word order seems to
arouse may help learners to monitor their production in order to eliminate the
effects of transfer. This might explain why Rutherford’s Japanese learners
displayed no evidence of the LI order in their written compositions.

Minimalist theoretical positions'on transfer

Minimalist theoretical positions on transfer have tended to emphasize the
similarity between L2 and LI acquisition. Newmark (1966), for instance,
drew extensively on research in LI acquisition in dismissing behaviourist ac¬
counts of L2 learning. He was primarily concerned with rejecting the view of
language as an incremental process, according to which learners were sup¬
posed to acquire a language structure by structure, proceeding from the sim¬
plest to the most complex. He acknowledged ‘interference’ but saw it as of
little importance, arguing that it simply reflected ignorance. Newmark and
Reibel (1968) produce the following description of what has become known
as the Ignorance Hypothesis:

... a person knows how to speak one language, say his native one; but in
the early stages of learning his new one, there are many things that he has
not yet learned to do ... What can he do other than use what he already
knows to make up for what he does not know? To an observer who knows
the target language, the learner will be seen to be stubbornly substituting
the native habits for target habits. But from the learner’s point of view, all
he is doing is the best he can: to fill in his gaps of training he refers for help
to what he already knows (1968: 159).
r
Krashen (1983) adopts a similar position, drawing directly on Newmark’s
ideas. He views transfer as ‘padding’, the result of falling back on old know¬
ledge when new knowledge is lacking. The cure is ‘comprehensible input’ (see
Chapter 7). In effect, Newmark and Reibel and Krashen treat LI transfer as a
kind of communication strategy (i.e. a means of overcoming a communica¬
tion problem) rather than as a learning strategy (i.e. a device for developing
interlanguage). This distinction between communication and learning trans¬
fer will be considered in greater detail later.
In place of the behaviourist notions of ‘habit formation’ and ‘interference’,
researchers such as Dulay and Burt (1972) have posited the existence of ‘gen¬
eral processing strategies’, which were seen as universal in the sense that they
were used by all language learners, first and second. They identified a number
 Language transfer 315

of general production strategies to account for the various types of errors they
observed. For example, the absence of grammatical functors was explained in
terms of ‘the pervasiveness of a syntactic generalization’ while the use of in¬
tonation to signal questions in the early stages of learning was the result of
‘using a minimal number of cues to signal the speaker’s intentions’. Occa¬
sional instances of transfer errors were explained away as the result of the
pressure to perform in the L2 (Newmark’s idea), of the nature of the learning
environment (for example, Ervin-Tripp (1974)) claimed that interference was
more likely in contexts where the L2 was not part of the learner’s larger social
milieu 4), and of the use of tasks, such as translation, that predisposed learners
to refer to their LI.
In retrospect, there can be little doubt that some scholars were too ready to
reject transfer as a major factor in L2 acquisition. This over-reaction was
caused by the close connection between the ideas of transfer and behaviour¬
ism, which, as we have seen, had become discredited. In clambering on to the
mentalist bandwagon, however, researchers like Dulay and Burt mistakenly
dismissed transfer, often on the basis of flimsy evidence. Subsequently, re¬
searchers have sought to relocate transfer within a cognitive framework and
have focused their endeavours on identifying the precise conditions that lead
to it.

Constraints on transfer
We have seen that a minimalist position on the role of transfer is not justified;
transfer constitutes an important factor in L2 acquisition. However, we have
also seen that behaviourist accounts of transfer, as reflected in the CAH in
particular, overpredict both the transferability of specific items (that is, they
fail to explain when they are transferred and when they are not), and transfer
load (how much is transferred). Increasingly, researchers have sought to
identify the conditions that promote and inhibit transfer. In this section we
will consider a number of differing constraints on transfer that incorporate
linguistic, psycholinguistic, and sociolinguistic factors. These constraints are:
(1) language level (phonology, lexis, grammar, and discourse), (2) social fac¬
tors (the effect of the addressee and of different learning contexts on transfer),
(3) markedness (the extent to which specific linguistic features are ‘special’ in
some way), (4) prototypicality (the extent to which a specific meaning of a
word is considered ‘core’ or ‘basic’ in relation to other meanings of the same
word), (5) language distance and psychotypology (the perceptions that
speakers have regarding the similarity and difference between languages),
and (6) developmental factors (constraints relating to the natural processes of
interlanguage development). It should be noted that these are not the only
constraining factors. Non-structural factors such as individual learner dif¬
ferences (such as personality and age), and the nature of the tasks a learner is
316 Internal factors

performing, also constrain LI transfer. However, as these are considered in

some detail elsewhere (see Chapters 11 and 5) they are not dealt with here.

1 Language level
There is widespread recognition that transfer is more pronounced at the level
of the sound system than at the level of syntax. The existence of ‘foreign ac¬
cents’ in L2 learning is so well attested that it hardly requires documenting. In
general, native speakers have little difficulty in distinguishing the language
background of different learners. Furthermore, as Purcell and Suter (1980)
have shown, the LI of learners serves as the best predictor of native speakers’
evaluations of their speech. In this study, the native speakers were asked to
judge the pronunciation accuracy of L2 learners of English with different Lis
(Thai, Japanese, Arabic, and Persian). The greater the difference between the
LI and the L2, the lower the rating. Thus, Thai and Japanese learners were
rated as less accurate than Arabic and Persian. However, the obvious effects
of the LI on pronunciation, while indicative of the importance of transfer at
this level, do not warrant uncritical acceptance of the ‘difference = difficulty’
hypothesis. As we shall soon see, learners do not invariably transfer the
phonological features of their LI. As is the case with syntax, phonological
transfer is governed in part by universal developmental tendencies (see Major
1986).
Similar evidence exists regarding the prevalence of transfer of LI lexis. Kel-
lerman comments ‘there are enormous quantities of evidence for the influence
of the LI on IL (interlanguage) when it comes to lexis’ (1987: 42). For ex¬
ample, Ringbom (1978) found that the majority of lexical errors made by
Swedish and Finnish learners of L2 English could be attributed to the transfer
of partial translation equivalents. Also, the acquisition of lexis appears to be
facilitated if the LI and L2 are related languages. Thus, Sjoholm (1976) re¬
ports that Swedish learners of L2 English did better in vocabulary learning
than Finnish learners, Swedish being closer than Finnish to English.
Th^re is also general acceptance that transfer is a major factor at the level of
discourse. In a frequently cited paper, Schachter and Rutherford (1979) argue
that what might at first sight appear to be transfer-induced syntactic errors
are, in fact, transfer-induced discourse errors. They examined errors of the
following type produced by Chinese and Japanese learners of L2 English:

Most of food which is served in such restaurant have cooked already.

Irrational emotions are bad but rational emotions must use for judging.

which were judged by native speakers to reflect a confusion between active

and passive. Schachter and Rutherford, however, claim that these sentences
reflect the transfer of the topic-comment structure found in Chinese and
Japanese. They suggest that learners learn a particular target language form
and then hypothesize that the form is used to express a particular discourse
 Language transfer 317

function. The topic-comment structure is, in fact, a universal feature of early

interlanguage (see Chapter 9, page 371), but where the LI supports its use—
as in the case of the Japanese and Chinese learners—it is more prevalent.
In a subsequent paper, Rutherford (1983) argues that whereas there is clear
evidence of transfer involving topic prominence and pragmatic word-order in
Japanese learners of English, there is no evidence of transfer involving
grammatical word order. He comments: ‘I take these observations as evid¬
ence that it is therefore discourse and not syntax that gives gross overall shape
to interlanguage’ (1983: 368). However, as we have seen in the previous sec¬
tion, Rutherford’s rejection of word order transfer is open to criticism. Thus,
while there can be few objections to his claim that discourse and pragmatic
transfer is common (see Kasper 1992 and Chapter 5 for evidence supporting
this), his assertion that it is more prevalent than syntactical transfer needs to
be treated with circumspection. It is not even clear how the relative occur¬
rences of transfer in discourse and syntax should be measured.5
It is, in fact, very difficult to quantify the extent of transfer in the different
language levels. There is, however, a theoretical reason for expecting the in¬
fluence of the LI to be felt more strongly in pronunciation, lexis, and dis¬
course than in syntax. In the previous section we noted Odlin’s (1990)
arguments that metalingual awareness may inhibit transfer in the case of
word order. It is probably true to say that most learners have a much more
highly developed metalingual awareness of grammatical properties than of
phonological or discourse/pragmatic properties. This awareness may enable
learners to control their choice of linguistic form at the level of grammar to a
greater extent than at the other language levels and this may inhibit transfer.

2 Sociolinguistic factors
Sociolinguistic factors have also been shown to influence when and to what
extent transfer takes place. We will consider the effects of (1) the social con¬
text and (2) the relationship between the speaker and the addressee on
transfer.
The social context can influence the extent to which transfer occurs. Odlin
(1989; 1990) has suggested that negative transfer is less likely in focused con¬
texts, where there is concern to maintain the standardness of languages, than
in unfocused contexts. This distinction between focused and unfocused con¬
texts is taken from Le Page and Tabouret-Keller (1985), who claim that
whereas some communities have a very clear idea of what constitutes a
language, others do not, mixing languages without much concern for what is
‘grammatical’ or ‘ungrammatical’. Odlin suggests that negative transfer is
less common in classroom settings than in natural settings because in the
former, learners constitute a ‘focused’ community and as a consequence treat
LI forms as intrusive and even stigmatized. In natural settings learners may
comprise either a ‘focused’ or an ‘unfocused’ community; where they are
318 Internal factors

unfocused, language mixing will be freely permitted, thus encouraging

negative transfer to take place.6 In addition, classroom learners are often
explicitly warned when interference might occur through the contrastive
presentation of items.
However, as we noted in Chapter 4, it may be misleading to talk of ‘com¬
munities’ where many L2 learners are concerned. It is likely that when learn¬
ers are in a classroom setting, they will adhere to target-language norms and
thus try to avoid negative transfer. However, when the same learners are out¬
side the classroom, they may show much less regard for target-language
forms and transfer quite freely.7 Thus, where L2 learners are concerned,
rather than talk of ‘communities’, it may be better to consider the effect of
social context in relation to the kind of norm—external or internal—that
learners have in mind. If the context requires attention to external norms (as
manifested in textbooks, reference books, and the teacher), negative transfer
is inhibited; if, however, the context encourages attention to internal norms
(as in free conversation involving speakers with shared languages), learners
may resort more freely to the LI if this helps comprehensibility and promotes
positive affective responses. Such variable transfer behaviour is most likely in
official language settings (see Chapter 6), where the L2 is used both inside and
outside of the classroom, but it may also be found in other settings as well. Of
course, learners with access to only one type of setting (such as the classroom)
may behave more like a ‘community’ where transfer is concerned.
Generalizations regarding the effect of macro contexts or settings on LI
transfer are dangerous, however, as they overlook the influence that specific
social factors can have on learners’ use of their LI. Studies by Beebe (1977)
and Beebe and Zuengler (1983) examined the effect of addressee factors on
transfer. Bilingual Thai/Chinese learners of English drew variably on their
two Lis depending on whether their interlocutor was Thai or Chinese. Beebe
(1980) found that Thai learners of English made use of a native variant of /r/
to a greater extent in a formal than in an informal context. Drawing on these
and other studies of variability, Tarone (1982) has argued that LI transfer is
likely to be more evident in learners’ careful style than in their vernacular
style, on the grounds that when learners are paying greater attention to how
they speak, they are more likely to make use of all their potential resources,
including LI knowledge. This variable use of the LI was considered in more
detail in Chapter 4.
It is interesting to note that the conclusion reached by Odlin on the basis of
a macro-sociolinguistic perspective is very different from that reached by
Tarone on a basis of a micro-sociolinguistic perspective. Given that ‘focused’
classroom learners are likely to make extensive use of a ‘careful style’, it seems
contradictory to claim that transfer is inhibited in such learners and yet also
prevalent in the careful style. This contradiction may be more apparent than
real, however. Classroom learners may indeed seek to avoid transfer in gen¬
eral recognition of the importance of external norms, but in cases where the
use of an LI norm appears socially appropriate (because, for example, it gives
 Language transfer 319

prestige) they may still resort to transfer. Clearly, though, much work needs
to be done to sort out such apparent contradictions.

3 Markedness

One of the strongest claims in recent research on transfer is that the transfer-
ability of different features depends on their degree of markedness. The term
‘marked’ has been defined in different ways, but underlying all of the defini¬
tions is the notion that some linguistic features are ‘special’ in relation to
others, which are more ‘basic’. Thus, for example, the adjectives ‘old’ and
‘young’ can be considered unmarked and marked respectively, because
whereas ‘old’ can be used to ask about a person’s age:
How old is she? (= what is her age?)

‘young’ cannot, except in some very special sense:

How young is she? (= is she as young as she makes out?).

More technical definitions of ‘markedness’ can be found in different lin¬

guistic traditions. These are discussed in some detail in Chapter 10, so here we
will provide only a brief account to enable us to examine the hypothesized re¬
lationship between markedness and transfer. One definition of ‘markedness’
derives from Chomsky’s theory of Universal Grammar. This distinguishes
the rules of a language that are core and periphery, as shown in Figure 8.1.
Core rules are those that can be arrived at through the application of general,
abstract principles of language structure, which Chomsky and other generat¬
ive linguists have held to be innate. Basic word order, for example, is con¬
sidered part of the core. Peripheral rules are rules that are not governed by
universal principles; they are idiosyncratic, reflecting their unique historical
origins. The structure ‘the more ... the more’ is an example of a peripheral
rule in English. Peripheral rules are marked. Core rules, as we will see in
Chapter 10, can be both unmarked and marked.

core

periphery

Figure 8.1: Markedness in core and peripheral grammar

Another definition of markedness is found in language typology (the study
of different types of language carried out in order to identify those properties
 320 Internal factors

that are universal). The identification of typological universals has been used
to make claims about which features are marked and which ones are
unmarked. The broad claim is that those features that are universal or present
in most languages are unmarked, while those that are specific to a particular
language or found in only a few languages are marked. Drawing on language
typology, Zobl (1984) offers three senses in which rules can be marked. The
first is typological specialization. For example, whereas English adheres to
the universal tendency of languages to avoid non-extractable ‘how’ in ‘how’ +
adjective phrases:

I didn’t realize how comfortable I was.

*1 didn’t realize how I was comfortable.

French permits extraction of ‘combien’ in equivalent sentences. The struc¬

tural properties of a language may also be marked as a result of typological in¬
consistency. Zobl cites the example of German and Dutch, which permit two
different word orders, one in main clauses (SVO) and the other in subordinate
clauses (SOV). Thus, German and Dutch word order can be considered
marked in relation to English word order, which displays a high level of con¬
sistency. Finally, typological indeterminacy occurs when a structure pre-
v dieted on the basis of a language’s overall typology is not found. For example,
English, as a SVO language, might be expected to manifest a noun + adjective
ordering, but does not do so. Also, some linguistic properties—such as adver¬
bial position—are not stringently controlled by overall typology and so can
be considered ‘fuzzy’. Typologically indeterminate features are seen as
marked.
There is somewhat mixed evidence regarding the effects of markedness on
the transferability of LI features. Two general hypotheses have been investig¬
ated: (1) learners will transfer unmarked forms when the corresponding TL
form is marked, and (2) learners will resist transferring marked forms, espe¬
cially when the corresponding TL form is unmarked. As Hyltenstam (1984)
puts it:

Unmarked categories from the native language are substituted for corres¬
ponding marked categories in the target language ... Marked structures
are seldom transferred, and if they are transferred, they are much more
easily eradicated from the target language (1984: 43).

Zobl has provided several examples of how learners tend to fall back on
their LI if the corresponding L2 rule is obscure because it is typologically in¬
consistent or indeterminate. For example, Zobl (1983a) notes that the fol¬
lowing errors are common in French learners of L2 English:

* They have policeman for stop the bus.

* He do that for to help the Indians.
 Language transfer 321

The ‘for’ + infinitive error corresponds closely to the LI structure (‘pour’ + in¬
finitive). Zobl points out that there are dialects of English where ‘for to’ oc¬
curs, that transformation grammar posits ‘for to’ as the deep structure and
that it appears in Old and Middle English. He suggests that this creates a
‘structural predisposition’ for transfer. In other words, ‘for to’ can be con¬
sidered unmarked in relation to the modern English structure. Zobl’s basic
point—reiterated in a number of papers (Zobl 1980a; 1980b; 1982; 1983a
and 1984)—is that an LI rule must meet certain conditions before it will be
transferred; it must be productive in the LI (not some kind of exception), it
must be used frequently, and it must not be ‘on the way out’ historically
speaking.
Evidence for the second hypothesis comes from Zobl (1984). On the
grounds that extraction is typologically specified (i.e. marked), Zobl predicts
that French learners of English will not accept ungrammatical English sen¬
tences such as the following:

* How many do you want oranges? (1984: 86),

even though such extraction is permitted in LI French. The results of a study

in which a grammaticality judgement task was administered to mixed profi¬
ciency learners lent broad support to this hypothesis. Interestingly, though,
the low-intermediate learners were more inclined to accept extraction than
either the beginner or advanced level learners (i.e. there was evidence of
U-shaped development involving transfer). Zobl suggests that this may reflect
the intermediate learners having noticed that English is tolerant of extraction
in other grammatical areas (for example, preposition stranding, as in ‘Who
did he give it to?’). This study suggests that (1) learners do resist transferring
marked forms when the corresponding TL structure is unmarked, but that (2)
this resistance can be overcome if learners obtain evidence that transfer is pos¬
sible. There is counter-evidence relating to (1), however. A study by Liceras
(1985) suggests that learners may be prepared to accept transfer of a marked
structure manifesting extraction. Liceras investigated preposition stranding
by English-speaking learners of L2 Spanish. English permits extraction of the
preposition in sentences like:

Who did John give the book to?

whereas Spanish does not. In this study, 43 per cent of the beginners accepted
stranding in Spanish. Liceras’ results, therefore, contradict Zobl’s.
Indeed, not all researchers take the view that learners will resist trans¬
ferring marked LI forms. White (1987b: 266-7) argues forcefully that ‘trans¬
fer is not confined to unmarked forms, that L2 learners may transfer marked
forms from the LI to the interlanguage, and that such transfer is compatible
with the theory of markedness currently invoked in generative grammar’. She
points out that the crucial test of the markedness hypothesis occurs only in
situations where the LI has a marked structure but the L2 does not. Such a
322 Internal factors

situation arises for English learners of L2 French in the case of double object
constructions, as in:

John gave Fred the book,

and preposition stranding, as in:

Who(m) did John give the book to?

neither of which occur in French. White found that whereas the English
learners she investigated provided evidence of resisting the transfer of the
marked preposition stranding construction in their grammaticality judge¬
ments, they showed a readiness to transfer the marked double object con¬
struction. Interestingly, another group of learners with different FI
backgrounds but with a knowledge of English as their first F2, were prepared
to transfer both structures. The learners with El English, therefore, behaved
anomalously, in that they alone resisted transferring one of the marked struc¬
tures. White suggests that one possible explanation for this exceptional
behaviour could be the influence of prescriptive English mother-tongue
teaching, where stranding is often presented as stylistically undesirable.
White concludes that overall her study supports the view that marked forms
are transferred and refers to other studies that show the same (for example,
Selinker, Swain, and Dumas 1975 and Tarallo and Myhill 1983).
Some of the most convincing evidence of markedness effects on transfer
can be found in studies that have examined asymmetrical patterns. A straight
contrastive analysis is unable to cope with evidence that shows that a given
feature (z) is transferred in one direction (i.e. transfer of z occurs from lan¬
guage x to language y) but not in the other (i.e. transfer of z does not occur
from language y to language x). A theory of transfer that incorporates mar¬
kedness, however, can provide an explanation for such phenomena.
The study that is most commonly cited to illustrate asymmetrical patterns
is Eckman (1977). Eckman investigated transfer in English learners of L2
German and German learners of L2 English, focusing on voice contrast in
pairs of phonemes such as III and /d/. In English this contrast exists word ini¬
tially (for example, ‘tin’ v. ‘din’), medially (for example, ‘betting’ v. ‘bed¬
ding’), and finally (for example, ‘wed’ v. ‘wet’). In German, however, the
distinction only exists word initially and word medially; in word-final posi¬
tion, only voiceless stops occur. Both the German and the English L2 learners,
therefore, are faced with learning to make a known distinction (i.e. voiced/
voiceless stops) in a new position. Eckman argues that typologically, voice
contrast in word-final position is more marked than in the other two posi¬
tions. He provides evidence to show that English learners have no difficulty in
learning that German has no voicing in word-final stops, but that German
learners experience considerable problems in learning that English does. In
other words, no transfer effects are evident when the LI position is marked
 Language transfer 323

and the L2 position marked, but they appear when the LI position is
unmarked and the L2 marked.
In a subsequent study, Eckman (1981) provides evidence to show that
devoicing of English stops in the most marked position (i.e. word finally) oc¬
curs in learners whose Lis, like English, include voicing. Cantonese learners,
for instance, sometimes say ‘pick’ for ‘pig’. This suggests, as Eckman acknow¬
ledges, that the devoicing of final stops is a ‘natural’ phenomenon. In the case
of the German learners, therefore, the transfer may be reinforcing what is a
universal tendency—a point we take up later when we consider develop¬
mental constraints on transfer.
In order to explain how markedness affects transfer, Eckman advances the
Markedness Differential Hypothesis:

Those areas of difficulty that a second language learner will have can be
predicted on the basis of a comparison of the native language (NL) and the
target language (TL) such that:
(a) those areas of the TL that are different from the NL and are relatively
more marked than in the NL will be difficult;
(b) the degree of difficulty associated with those aspects of the TL that
are different and more marked than in the NL corresponds to the rel¬
ative degree of markedness associated with those aspects;
(c) those areas of the TL that are different from the NL but are not
relatively more marked than the NL will not be difficult.
(Eckman 1977: 321)

The notion of ‘markedness’ is defined typologically; that is, an area (X) is to

be considered relatively more marked than some other area (Y), if cross-
linguistically X implies the presence of Y, but Y does not imply the presence of
X. An example is found in the comparison of voiced and voiceless stops in
English and German discussed above. The Markedness Differential Hypo¬
thesis constitutes an attempt to reformulate the CAH to take account of
markedness factors. As Eckman (1985) points out, it differs from the CAH in
a number of important ways: it seeks to explain (1) not only where learning
difficulty will occur, but also the relative degree of difficulty; (2) where differ¬
ences between the native and target languages will not result in difficulty; and
(3) why certain structures are typically acquired before other structures. As
such it constitutes a much more powerful theory.
One problem with work involving markedness is the vagueness of the con¬
cept. Eckman (1985: 306) acknowledges this by admitting that ‘one area in
which more research is needed is in defining markedness relations’. The con¬
cept is characterized by a fuzziness that sometimes makes it difficult to deter¬
mine which features are marked in relation to others. This is, in part, the
result of differences in the linguistic descriptions and shifts in the linguistic
theories upon which claims regarding the degree of markedness of specific lin¬
guistic properties rest. One way in which greater precision can be given to the
324 Internal factors

concept is by defining it with reference to ‘native speakers’ own perceptions of

the structure of their language’ (Kellerman 1977). Thus, instead of de¬
termining markedness through reference to some linguistic description or
theory, as was the case in the studies discussed in this section, it can be deter¬
mined by asking native speakers whether they perceive specific features as ‘in¬
frequent, irregular, semantically or structurally opaque, or in any other way
exceptional’ (Kellerman 1983: 117). We now turn to Kellerman’s work on
markedness, or, as he calls it, prototypicality.

4 Prototypicality
In a series of papers, Kellerman (1977; 1978; 1979; 1986; 1989) sought to
demonstrate that learners have perceptions of the structure of their own lan¬
guage, treating some structures as potentially non-transferable and others as
potentially transferable, and that these perceptions influence what they actu¬
ally transfer. The maj ority of the studies that Kellerman carried out to test this
hypothesis have examined lexico-semantics. Unlike the research reported in
the previous section, which was based on analyses of learner language, Keller¬
man’s studies make use of native speakers’ intuitions regarding their LI.
The best known of Kellerman’s studies is the ‘breken’ study (Kellerman
1978). This study had two stages. In the first stage, native speakers of Dutch
were asked to sort seventeen sentences containing the verb ‘breken’ (see Table
8.4) into groups so that the sentences in each group were similar in meaning.
Kellerman then examined the number of times a given pair of sentences were
placed in the same group. A multidimensional scaling analysis was carried
out to investigate the ‘dimensions’ of the ‘semantic space’ occupied by ‘bre¬
ken’. Kellerman (1979) reports two major dimensions, which he labelled
‘core/non-core’ and ‘concrete/abstract’. In the second stage of the study, Kel¬
lerman asked 81 Dutch students of English in their first and third years at uni¬
versity to say which of the 17 sentences containing ‘breken’ they would
translate using the English verb ‘break’. There were clear differences in the
percentage of students prepared to translate each sentence. For example,
whereas 81 per cent considered ‘hij brak zijn been’ (= he broke his leg) trans¬
latable only 9 per cent identified ‘sommige arbeiders hebben de staking gebro-
ken’ (= some workers have broken the strike) as translatable. Kellerman
found that the rank order for the ‘transferability’ of the 17 sentences correl¬
ated poorly with the ‘concrete/abstract’ rank order derived from native
speaker intuitions. However, it correlated strongly and significantly with the
‘core/non-core’ order. This led Kellerman (1979: 51) to conclude that native
speakers’ intuitions about semantic space can be used to predict transferabil¬
ity, at least for the meanings of ‘breken’ and also that ‘the most important fac¬
tor is “coreness”.’ It should be noted that Kellerman’s results could not have
been predicted on the basis of a contrastive analysis, as the learners clearly
resisted transferring ‘brekens’ that had exact equivalents in English.
 Language transfer 325

Dutch English
1 Hij brak zijn been. He broke his leg.
2 ’t Kopje brak. The cup broke.
3 Na’t ongeluk is hij ’n gebroken man After the accident, he became a broken
geworden. man.
4 Zij brak zijn hart. She broke his heart.
5 De golven braken op de rotsen. The waves broke on the rocks.
6 De lichtstralen breken in het water. The light rays refract in the water.
7 Dankzij ’n paar grapjes, was’t ijs eindelijk Thanks to a couple of jokes, the ice was
gebroken. finally broken.
8 Hij brak zijn woord. He broke his word.
9 De man brak zijn eed. The man broke his oath.
10 ‘Nood breekt wet.’ ‘Necessity breaks law’ (a saying).
11 Zij brak’t wereldrecord. She broke the world record.
12 Zijn val werd door ’n boom gebroken. His fall was broken by a tree.
13 Zijn stem braktoen hij 13 was. His voice broke when he was 13.
14 Sommige arbeiders hebben de staking Some workers have broken the strike.
gebroken.
15 Welk land heeft de wapenstilstand Which country has broken the ceasefire?
gebroken?
16 Het ondergrondse verzet werd gebroken. The underground resistance was broken.
17 n’ spelletje zou de middag enigszins A game would break up the afternoon a
breken. bit.

Table 8.4: Sentences with ‘breken’ ranked according to coreness (prototypicality)

(adapted from Kellerman 1979: 49)

In an extension of the ‘breken’ study, Kellerman (1979) asked 291 learners

of English (including the 81 subjects in the study described above), who
ranged from twelve-year-olds in their second year of English to third year
university students, to assess the translatability of nine of the 17 ‘breken’ sen¬
tences. He found that the rank orders for the different groups of learners were
‘remarkably consistent’ and concluded that ‘the effects of teaching, learning
and growing older do not significantly alter learners’ beliefs about the relative
transferability of the “brekens”’ (1979: 52). Kellerman’s study, therefore,
suggests that learners’ perceptions of what is transferable are not influenced
by their L2 proficiency, but clearly this needs further investigation.
The concepts of ‘coreness’ and ‘markedness’ are obviously related, which
Kellerman (1983) acknowledges. In this later paper, he refers to ‘psycho-
linguistic markedness’ in recognition of the importance he attaches to native
speakers’ perceptions of the structure of their own language. In a further
terminological switch, Kellerman (1986) refers to prototypicality to label the
 326 Internal factors

same basic idea. He draws on work on ‘semantic prototypes’ in cognitive psy¬

chology, citing Coleman and Kay (1981):

a semantic prototype associates a word or phrase with a prelinguistic cog¬

nitive schema or image ... speakers are equipped with an ability to judge
the degree with which an object (or ... the internal representation thereof)
matches this prototype schema or image.

Kellerman suggests that the prototypical meaning of a lexical item, such as

‘breken’, is that which a dictionary gives as the primary meaning of the item.
The bulk of Kellerman’s empirical work has concerned lexico-semant-
ics—the early ‘breken’ study and the later ‘eye’ study (Kellerman 1986), for
example. However, one study has also examined a syntactical structure, con¬
ditionals. Kellerman (1989) provides evidence to show that advanced Dutch
learners of L2 English are likely to produce errors of this kind:

* If it would rain, they would cancel the concert in Damrosch Park ( = If it

rained they would cancel the concert in Damrosch Park).

despite the fact that Dutch makes use of equivalent verb forms in main and
subordinate clauses to English. Kellerman suggests that the failure to transfer
v the Dutch verb forms is the result of two tendencies. One is the learners’ res¬
istance to transferring a marked form. In this case, markedness is associated
with the idea of a ‘semantically transparent grammar’. It is more transparent
to say ‘would rain’ than ‘rained’ because the verb is explicitly marked for fu¬
ture time. The second tendency is that of symmetry—the attempt to match
the verb forms in the main and subordinate clauses. Kellerman points out that
the same tendencies are also evident in both standard and non-standard vari¬
eties of the target language (English), suggesting that there is ‘an interaction
between natural tendencies and the native language’ (1989: 111). This, he
speculates, may explain why fossilization occurs.
A number of points emerge from Kellerman’s work on prototypicality. The
first is that it is possible to provide a clear operational definition of ‘marked¬
ness’ or, ‘prototypicality’ by making use of native speakers’ judgements of
‘similarity’. The second is that learners have perceptions about what is trans¬
ferable from their LI and act in accordance with these perceptions. The third
is that these perceptions reflect learners’ ideas about what is prototypical or
semantically transparent in their LI. Kellerman (1983) suggests that learners
prize ‘reasonableness in language’ and ‘attempt to keep their L2s transparent’
(1983: 129). LI structures that they perceive to be working against this prin¬
ciple—such as idioms that are highly metaphorical or grammatical structures
where meanings are not overtly encoded—are not transferred. Finally, Kel¬
lerman has shown that learners’ perceptions regarding the translatability of
LI items is not influenced by their experience with the L2.
While Kellerman’s work has gone a long way to teasing out the nature of
the constraints on positive transfer, a few words of caution are in order. With
 Language transfer 327

the exception of the study on conditionals, Kellerman’s research has been

based on the elicitation of native speakers’ intuitions regarding the similarity
and translatability of decontextualized sentences. Kellerman (1986) justifies
this by pointing out that corpora of spontaneous speech are unlikely to supply
the crucial data needed to test hypotheses. The weakness of such an ap¬
proach, which Kellerman acknowledges, is that we do not know to what ex¬
tent learners’ judgements about what can be done accurately reflect what they
actually do when using the L2. It is perhaps also wise to exercise caution
about equating ‘translatability’ with ‘transferability’. In Chapter 4 we exam¬
ined a number of studies which show that differences in LI transfer occur de¬
pending on whether the task requires translation or some other kind of
performance. It does not follow that learners are prepared to make the same
use of their LI in natural speech as they are in translation.
Kellerman’s research raises an important question; what kind and how
much evidence do learners need before they will accept non-prototypical ele¬
ments, present in their LI, into their interlanguages? This is a crucial question
because it concerns how learners overcome the kind of constraints on positive
transfer that Kellerman has identified. However, as Kellerman (1986) notes
there are, as yet, no clear answers.

5 Language distance and psychotypology

We now turn to consider another constraint on LI transfer—the ‘distance’
between the native and the target languages. Distance can be viewed as both a
linguistic phenomenon (i.e. by establishing the degree of actual linguistic dif¬
ference between two languages) or a psycholinguistic phenomenon (i.e. by
determining what learners think is the degree of difference between their nat¬
ive language and the target language). Kellerman (1977) used the term psy¬
chotypology to refer to learners’ perceptions about language distance.
There is substantial evidence to indicate that the actual distance between
the native and the target languages acts as a constraint on transfer. The im¬
portance of this factor is reflected in the different amounts of time which the
Foreign Service Institute in the United States allocates to courses aimed
at achieving a high level of proficiency in different languages (for example,
20 weeks for French as opposed to 44 for Serbo-Croatian). Similarly, the Brit¬
ish Foreign Service pays different rates of language proficiency allowance
according to a scale of difficulty in learning different groups of languages.
Language distance can affect L2 learning either through positive transfer or
through negative transfer. Corder (1978b; 1981a: 101) chooses to emphasize
positive transfer, arguing:

... other things being equal (e.g. motivation and access to data etc.), the
mother tongue acts differentially as a facilitating agency. Where the mother
tongue is formally similar to the target language the learner will pass more
328 Internal factors

rapidly along the developmental continuum (or some parts of it), than
where it differs.

It is likely, however, that language distance is a factor in both positive and

negative transfer.
Evidence for this claim comes from the research conducted in Finland on
the acquisition of English by Swedish-speaking and Finnish-speaking Finns
(see Sjoholm 1979; Ringbom 1976; 1978; and 1987). Finland constitutes an
ideal setting for testing the effects of language distance. Whereas 93 per cent
of the population speak Finnish as their mother tongue, a language distant
from English, 6 to 7 per cent speak Swedish, a language much closer to Eng¬
lish. Both groups consider themselves to belong to the same culture, however.
Sjoholm, Ringbom, and their co-researchers have been able to show that
Swedish-speaking Finns enjoy a substantial learning advantage over Finnish¬
speaking Finns. For example, Sundquist (1986) found that Swedish-speaking
Finns did as well in reading comprehension tests after one and a half years as
Finnish-speaking Finns did after three and a half. Interestingly, however,
there is evidence to suggest that, contrary to the predictions of the CAH, the
Finnish-speaking Finns often make fewer errors, i.e. they manifest less nega¬
tive transfer. Sjoholm (1976), for instance, found fewer Fl-based errors in
Finnish-speaking university students than in the Swedish speakers. Further¬
more, Ringbom (1978) was able to show that both the Swedish- and Finnish¬
speaking groups (both of whom are bilingual in the two languages) were
much more likely to transfer word morphology from Swedish than from
Finnish. Fanguage switch was entirely from Swedish and word blends were
almost always Swedish-English. In other words, whereas the Finnish speakers
avoided transferring elements from their FI, preferring instead to fall back on
their first F2 (Swedish), the Swedish speakers did transfer elements from their
FI but avoided doing so from their first F2 (Finnish).
Kellerman (1977) has claimed that learners possess a psychotypology (a set
of perceptions about language distance), and that it is this—rather than ac¬
tual distance—that triggers or constrains transfer. Fearners form ‘projec¬
tions’ about what can be transferred on the basis of their beliefs as to whether
the native and target languages are the ‘same’—either in terms of ‘linguistic
detail’ or ‘in very general terms’. On the basis of these projections, learning
decisions, or ‘conversions’, are made.
An extension of the ‘breken’ study reported in the previous section pro¬
vides support for the idea of a psychotypology. Kellerman compared Dutch
learners’ judgements regarding the translatability of the ‘brekens’ into F2
German (a language close to Dutch) with their judgements regarding their
translatability into F2 English (a language more distant from Dutch). The res¬
ults showed that, in general, whereas they accepted the sentences in German,
they sometimes rejected them in English. These results demonstrate a clear ef¬
fect for perceived language distance.
 Language transfer 329

Further evidence for the role of pyschotypology comes from Singleton’s

(1987) case study of Philip, an English-speaking learner of French. Philip dis¬
played a high level of communicative efficiency in French, despite the fact that
he had minimal opportunities to learn it. Singleton provides evidence to show
that he borrowed extensively from his known languages (Fatin, Spanish, and
Irish, as well as English). Philip utilized Romance sources (i.e. those that were
close to French) and, furthermore, was often able to attribute the forms he
borrowed to a particular language. He demonstrated well-informed notions
about which languages would assist him most in learning French. It would
seem, then, that learners do have clearly defined perceptions regarding the
similarities and differences between languages.
Kellerman (1979) argues that learners’ psychotypology is not fixed.
Rather, it is revised as they obtain more information about the target lan¬
guage. Thus, Dutch learners of German may start out with the assumption
that the target language is very similar to their mother tongue, but later on
come to adjust this perception as they recognize the many differences. Keller¬
man comments:

Thus experience affects the provisional typology the learner is building up.
This means that at any given moment certain NF (native language) features
will be available for transfer to the given TF (target language), and others
will not be (1979: 40).

It follows that FI items are not perceived as inherently ‘neutral’ (and so avail¬
able for transfer) or ‘specific’ (and so not available for transfer).
According to Kellerman, learners’ psychotypologies interact with their in¬
tuitive feel for prototypicality, which, it should be remembered, does not ap¬
pear to change with developing proficiency. Prototypicality determines what
learners are prepared to risk transferring. Their psychotypology determines
what is actually transferred in performance. On the basis of the perceived dis¬
tance between the native and target languages, learners decide whether to go
ahead and transfer those items that they perceive to be prototypical and,
therefore, potentially transferable. This interaction between psychotypology
and prototypicality results in an extremely complex process, especially as
learners’ psychotypologies change with experience.

6 Developmental factors
The constraints that developmental factors impose on FI transfer will be con¬
sidered with reference to (1) the extent to which transfer is evident at different
levels of development, and (2) the complex interplay between natural
principles of F2 acquisition and transfer.
330 Internal factors

The learner’s general level of development

One way of looking at interlanguage is as a restructuring continuum (Corder

1978a). That is, the starting point of L2 acquisition is the learner’s LI, which
is gradually replaced by the target language as acquisition proceeds. Such a
view suggests that transfer will be more evident in the early than the later
stages of development. This is what Taylor (1975) found when he compared
the proportions of transfer errors and developmental errors (such as overgen¬
eralization) produced by Spanish-speaking students of L2 English. The stu¬
dents in less advanced classes were more likely to make translation errors that
reflected their LI than were students in more advanced classes, who produced
more overgeneralization errors.
Further support for a relationship between the learner’s general level of de¬
velopment and the presence or absence of transfer comes from studies of L2
phonology. Major (1986), for instance, found that phonological transfer was
especially evident in the earlier stages of development. Wenk’s (1986) study
of French learners’ acquisition of L2 English rhythm also lends support to a
restructuring view of interlanguage. Wenk characterizes French rhythm as
‘trailer-timed’ and English as ‘leader-timed’. In the former, the accented syl¬
lable is towards the end of a rhythmic group, it is lengthened, and there is no
increase in loudness; unaccented syllables are relatively tense and vowels are
only weakly centralized. In the latter, the accented syllable is towards the be¬
ginning of a rhythmic group, it manifests only variable lengthening, but there
is an increase in loudness; in accented syllables the vowels are lax but strongly
centralized. Figure 8.2 shows the three stages of development that Wenk
found in French-English interlanguage. Beginner learners simply transfer
trailer-timed rhythm into English. Intermediate learners seem to produce a
kind of hybrid rhythm, with features of both trailer-timing and leader-timing.
Advanced learners exhibit standard English leader-timed rhythm. Wenk
points out the crucial items are cognates like ‘Japan’, ‘command’, and ‘po¬
lice’, where the accent falls on the final syllable in both the LI and L2. Such
items would seem to lend themselves to transfer, but Wenk found that they
were among the items judged least ‘native’ in a group of intermediate learners
and argues that this reflects the ‘transitional rhythmic grouping’ found at
stage 2.
However, although such evidence strongly suggests that learners gradually
restructure their interlanguage by replacing LI features with L2 features, cau¬
tion is needed. Not all errors in early interlanguage are traceable to trans¬
fer—many are intralingual and resemble those found in LI acquisition. This
is as true for phonology as it is for grammar (see Wode’s (1980) account of the
acquisition of L2 English phonology by four German children discussed
below).
Also, some errors, clearly attributable to LI influence, only emerge at later
stages of development. Kellerman (1983), for example, points out that trans-
 Language transfer 331

Figure 8.2: Three stages of development in the acquisition ofL2 English rhythm by
French learners (Wenk 1986)

fer errors involving pronominal copies in relative clauses can only occur when
the learner is at a sufficiently advanced stage of development to produce relat¬
ive clauses. Klein (1986: 27) also argues that ‘the possibilities of transfer in¬
crease as knowledge of the second language increases’. Bhardwaj (1986)
illustrates this in a detailed study of an adult learner of L2 English. This learn¬
er was unable to express his Punjabi conception of location until he had ac¬
quired some idea of the meaning of words like ‘up’, ‘down’ and ‘on’, and had
also developed the linguistic means to make these elements the head of defin¬
ite noun phrases. Only then did the influence of LI Punjabi appear in phrases
like ‘the up’, ‘the down’, etc.
Nor can it be assumed that transfer errors which appear at an early stage of
development are subsequently eliminated. Some transfer errors that appear at
an initial stage continue to manifest themselves in advanced learners. Bohn
and Flege (1992), for instance, found that German learners of L2 English
failed to develop target-like categories for vowels that were similar but not
332 Internal factors

identical to vowels in their LI, perhaps because ‘category formation is

blocked by equivalence classification’ (1992: 156). Finally, as Odlin (1989)
points out in his discussion of Taylor’s study, it is necessary to consider the
facilitating effects of transfer as well as negative transfer. It seems reasonable
to assume that advanced learners will be in a better position to take advantage
of similar LI material—such as cognate vocabulary—than beginners.
Interlanguage is clearly not a restructuring continuum. Although some as¬
pects of L2 development, such as rhythm, may reflect the gradual replacement
of LI by target-language features, other aspects do not. In some cases, trans¬
fer is only evident in the later stages of development, while in others early
transfer is never eliminated.

Natural principles of language acquisition

There is growing evidence to suggest that the LI and developmental factors
work together in determining the course of interlanguage—or, to put it an¬
other way, ‘transfer is selective along the developmental axis’ (Zobl 1980a).
This selectivity is evident in a number of ways: (1) the effects of the LI only
become evident when the learner has reached a stage of development that
makes transfer possible, (2) development may be retarded when a universal
transitional structure arising naturally in early interlanguage corresponds to
an LI structure, and (3) development may be accelerated when an early trans¬
itional structure is not reinforced by the corresponding L2 structure.
A number of studies show that the influence of the LI is developmentally
constrained in the sense that it only occurs when the learner has reached a
stage of development that provides a ‘crucial similarity measure’ (Wode
1976). Wode (1976; 1978; 1980) illustrates how this works. In the case of
negation, for instance, the children that he studied initially manifested the
universal pattern of development (see Chapter 3), but when they learnt that
the negative particle could follow the verb ‘be’ or an auxiliary/modal verb in
English, as in German, they assumed that it could also follow a main verb, as
it does in German but not in English. In other words, when confronted with
evidence that L2 negation worked in the same way as LI negation, they as¬
sumed that the two languages were completely identical in this structure. A
similar pattern of development has been observed in the acquisition of L2
English by other learners whose LI possesses post-verbal negation (for ex¬
ample, Norwegian—see Ravem 1968), but has not been observed in learners
whose Lis have other ways of conveying negation—for example, Japanese
(Milon 1974), Chinese (Huang 1970), or Spanish (Cazden et al. 1975). Wode
(1978) found further evidence for developmentally constrained LI transfer in
interrogatives. He concluded that L2 learners make use of LI syntactical
knowledge in systematic ways, depending on the formal properties of the
structures involved. He admits, though, that the exact nature of the ‘crucial
similarity measure’ is not clear.
 Language transfer 333

The notion of ‘developmental transfer’ is also applicable to L2 phonology.

Wode (1980) found that L2 phonological systems are acquired ‘through the
grid of the learner’s LI system’ (1980:129). Thus, elements of the L2 that are
sufficiently similar to elements in the learner’s LI repertoire will be substi¬
tuted by LI elements to begin with. However, L2 elements that fall outside
these ‘crucial similarity measures’ are not replaced with LI elements, but in¬
stead undergo autonomous development, similar to that observed in LI ac¬
quisition. Wode also notes that some elements appear to pose no learning
difficulty even when there is no equivalent LI element. What is missing is pre¬
cise information about the conditions under which LI transfer is activated.
As Leather and James (1991) conclude in their detailed survey of studies of
L2 speech acquisition, only ‘preliminary results’ are available.
There is also the possibility that errors can be doubly determined—that is,
reflect both ‘naturalness’ factors and LI influence. Hatch (1983a) identifies a
number of ‘naturalness’ factors that are independent of the LI, such as how
salient a feature is to the L2 learner and how transparent the relationship be¬
tween a particular form and meaning is. In a similar vein, Andersen (1983a)
also advances the view that:

Transfer can only function in conjunction with operating principles that

guide language learners and users in their choice of linguistic forms to ex¬
press the intended meaning (1983a: 180).

These principles (considered in greater detail in the next chapter) affect learn¬
ing difficulty and, thereby, the order and sequence of development in both LI
and L2 acquisition.
Hatch argues that both naturalness and LI transfer are at work in phono¬
logy and morphology, but that at higher levels of language—syntax and dis¬
course—naturalness factors may predominate. Some of the clearest examples
of ‘doubly-determined’ errors, however, are to be found in L2 syntax. Cazden
et al. (1975) noted that the early ‘no + verb’ negatives observed in the speech
of their Spanish-speaking learners cannot be attributed to LI transfer, even
though Spanish has the same negative pattern. This is because such negatives
are found in all learners, irrespective of the pattern in their LI. Cazden et al.
also noted that the ‘no + verb’ pattern persisted in the interlanguage of their
learners longer than was the case with other learners whose Lis did not have
this pattern. They concluded that when a developmental pattern coincides
with an LI pattern, progress may be hindered.
Finally, the LI has also been shown to have a facilitative effect when there
is a lack of correspondence between the LI pattern and a natural develop¬
mental pattern. Hammarberg (1979) re-analyses the data from Hyltenstam
(1977) to show that, although the original conclusion (i.e. learners with dif¬
ferent language backgrounds go through the same stages of development in
acquiring Swedish negation) holds true, some learners progressed through
these stages more rapidly than others and that this could be explained by their
334 Internal factors

language background. Thus, English-speaking learners, whose LI has post¬

verb negation, appeared to miss out the first stage (pre-verb negation). As Kel-
lerman (1987) puts it, the LI can provide ‘a leg-up along the developmental
ladder’.
It is clear that an acceptable theory of transfer must take account of how
learners’ previous LI knowledge interacts with the linguistic and cognitive
principles responsible for the universal properties of interlanguage develop¬
ment. The relative strength of contributions from these two sources also
needs to be determined. It is probably true to say that current work in transfer
treats the linguistic and cognitive principles as primary and LI knowledge as
secondary. Gass (1983), for instance, concludes her study of pronoun reten¬
tion (see page 303) by stating that ‘in considering the relationship between
NL facts and language universals, the latter were found to play the leading
role’ (1983: 79). However, such a conclusion may not be valid for all levels of
language, nor even for all aspects of syntax. Certainly, where phonology is
concerned, there are grounds for considering LI knowledge as primary, as
suggested by Wode’s and Wenk’s studies.
A theory of transfer must also specify in precise terms how and when the
two sources interact. One attempt to do this is Andersen (1983a). Using data
from a number of earlier studies, he tests out a number of constraints on
transfer, including all those considered above. He concludes that although it
is clear that transfer interacts with natural acquisitional principles, there is
still ‘leakage’ (i.e. it is not possible to fully predict when, how, and to what ex¬
tent transfer will take place). We are still a long way from a fully ‘develop¬
mental’ theory of transfer.

Summary

We have now examined six general constraints on LI transfer:

1 Language level

Although it is difficult to quantify the extent of transfer in the different lan¬

guage levels, there are some grounds for claiming that transfer is more con¬
spicuous at the levels of phonology, lexis, and discourse than grammar,
possibly because learners have a more developed metalingual awareness of
grammar.

2 Sociolinguistic factors

It has been suggested that when learners attend to external norms, as they are
likely to in classroom settings, transfer will be impeded. However, learners
may also make use of LI forms in their careful style if they have a strong social
motivation to do so.
 Language transfer 335

3 Markedness

There is some evidence to suggest that learners are ready to transfer

unmarked LI forms, but resist transferring LI marked forms; however, not
all the research points clearly in this direction. Eckman’s Markedness Differ¬
ential Hypothesis constitutes an improvement on the Contrastive Analysis
Hypothesis by indicating how markedness interacts with linguistic difference
to determine when transfer will and will not take place. Markedness, how¬
ever, remains a somewhat ill-defined concept.

4 Prototypicality

Kellerman tapped native speakers’ intuitions to determine which meanings of

a lexical item are unmarked or ‘prototypical’. Learners resist transferring
non-prototypical meanings.

5 Language distance and psychotypology

The actual distance between languages affects positive transfer; learners find
it easier to learn an L2 that is similar to their own language. However, per¬
ceived distance may be more important than actual distance. Kellerman sug¬
gests that whereas prototypicality influences what learners are prepared to
risk transferring, their psychotypology (which changes as their proficiency
develops) governs what they actually transfer.

6 Developmental factors
Whereas some researchers (for example, Taylor 1975) have claimed that neg¬
ative transfer is more evident in beginners, other researchers have argued that
learners may need to reach a certain stage of development before transfer of
some LI properties becomes possible. In general, except possibly where
phonology is concerned, the evidence does not support the claim that inter¬
language constitutes a restructuring continuum. Transfer interacts with
natural principles of L2 acquisition, sometimes occurring early on and some¬
times later. It can both retard and accelerate natural development.
The research that has investigated these constraints has helped us to under¬
stand what is transferred and what is not and also when transfer takes place
and when it does not. However, as yet there is no overall theory of transfer
that can account for how these various constraints interact.

Towards a theory of first language transfer

It should be clear from the preceding section that a theory of transfer is likely
to also be a general theory of L2 acquisition, in that the role of the LI cannot
easily be separated from other factors that influence development. The
336 Internal factors

theory, then, must account for how LI knowledge interacts with input in
shaping the learner’s interlanguage system and also how both LI and
interlanguage knowledge are drawn on in L2 production.
An important distinction—not always made in discussions of transfer—is
between transfer in L2 communication and transfer in L2 learning. This
distinction mirrors the distinction we made in Chapter 7 between the role that
input plays in comprehension and the role it plays in acquisition. There we
noted that input that works for comprehension may not always work for
acquisition. According to Kasper (1984b) and Fserch and Kasper (1986b),
transfer in communication involves the use of the LI to either receive
incoming messages (reception) or to process output (production). Transfer in
learning occurs when the learner uses the LI in the attempt to develop
hypotheses about L2 rules. There are, therefore, a number of possibilities; one
is that transfer is primarily a characteristic of communication, a second is that
it is primarily a feature of learning (i.e. learners draw directly on their LI in
constructing their interlanguages), while a third is that both communi¬
cation and learning transfer are significant and inter-related aspects of L2
acquisition.

Communication transfer

There is general acceptance that transfer in communication is common. Min¬

imalist accounts of transfer acknowledge the role played by the LI in com¬
munication while denying any role or any substantial role for the LI in
interlanguage hypothesis construction (i.e. in learning). Corder’s (1983) view
of transfer is that it is primarily a communication strategy, which he terms
borrowing. He emphasizes that borrowing is ‘a performance phenomenon,
not a learning process, a feature, therefore, of language use and not of lan¬
guage structure’ (1983: 92). It is invoked in order to compensate for defi¬
ciencies in the interlanguage system. He suggests that the term ‘transfer’ is, in
fact, inappropriate for this use of the LI as ‘nothing is being transferred from
anywhere to anywhere’. Corder acknowledges that the occurrence of bor¬
rowing is not the same for all learners, but is rather ‘a feature of the percep¬
tion of the relationship between first and second languages’ (1983: 93).
It should be noted, however, that not all communication transfer need be
strategic in nature. Fserch and Kasper (1986b; 1989) distinguish three types
of production transfer.8 In the case of strategic transfer, the learner gives focal
attention to a planning problem and to its solution, which may involve the use
of the LI. This is Corder’s ‘borrowing’. Subsidiary transfer occurs when there
is no focal awareness of either the production problem or of the transferred
LI knowledge, although awareness may develop later as a result of
monitoring. Automatic transfer takes place when the learner makes use of a
highly automatized LI subroutine. In this case, attention is completely
diverted to other aspects in the production process. Fserch and Kasper (1989)
 Language transfer 337

illustrate how these different types of production transfer can be distin¬

guished by studying ‘signals of uncertainty’ (for example, pausing, sighing,
laughing).
Communication transfer involves both production and comprehension
transfer. The former has received more attention than the latter, perhaps be¬
cause it is easier to identify in learner performance. However, Ringbom
(1992) has pointed out that ‘transfer is at least as important in comprehen¬
sion as it is in production’ (1992: 88). He provides evidence from Swedish¬
speaking Finns and Finnish-speaking Finns to show that the advantage which
the former derive from the closer proximity of their LI to English enables
them to outperform the latter in both reading and listening comprehension.
He suggests that the LI constitutes ‘potential knowledge’ that can be drawn
on more easily in decoding, which involves form-to-function mapping, than
in encoding, which involves function-to-form mapping. Ringbom also ac¬
knowledges constraints on transfer in both comprehension and production,
arguing that in both ‘the closer the perceived distance, the more (positive) the
transfer is’ (1992:106).
Transfer in communication is motivated by the learner’s desire to compre¬
hend or produce messages, but it may also have an effect on the process of hy¬
pothesis construction and testing, which many scholars see as central to
interlanguage development. In other words, transfer in communication may
lead to transfer in learning. Corder (1983) explains how this can happen:
... persistent communicatively successful borrowing works backwards, as
it were, and the successfully borrowed forms are eventually incorporated
into the interlanguage grammar, both the correct and the incorrect (1983:
94).
Ringbom (1992) suggests that it is transfer in comprehension that is most
likely to induce a change in the learner’s mental grammar. Such a view ac¬
cords with theories that give centrality to comprehensible input in L2 acquisi¬
tion. While acknowledging the legitimacy of Ringbom’s claim, we should
also recognize, as claimed by the comprehensible output hypothesis (see
Chapter 7), that L2 output and, therefore, transfer in production can also
contribute significantly to interlanguage development.

Learning transfer
Researchers like Corder seek to explain transfer entirely in terms of commun¬
ication; it is either a performance phenomenon or it is learnt as a product of
repeated performance. They reject the idea that learners transfer directly
from their LI into their interlanguages. Such a position is difficult to main¬
tain, however. For one thing, there is Evidence that particular transfer errors
occur in whole populations with the same LI. It is far-fetched to suggest that
all these learners engaged persistently in borrowing and as a result learnt the
LI structure. It is also not clear how communication transfer can account for
338 Internal factors

the fossilization of certain LI influenced structures in learners’ interlanguages

of the kind that Kellerman (1989) identified in advanced Dutch learners of
English.9
Thus, whereas some instances of transfer can be put down to the use of a
communication strategy, it is also necessary to recognize a more direct role
for the LI in L2 acquisition (i.e. transfer in learning). The general view, well
represented by Schachter’s (1983) ‘new account of transfer’, is that direct
learning transfer regularly occurs and that it can best be explained within a
cognitive rather than in a behaviourist framework. Transfer is conceptualized
as one ‘strategy’ operating within a general process of hypothesis construc¬
tion and testing.
Schachter argues that learners construct and reconstruct hypotheses by
means of inductive inferencing (scanning data, observing regularities, and
generalizing) and deductive inferencing (testing hypotheses by looking in the
first instance for confirming evidence and subsequently for disconfirming
evidence). She suggests that the learner begins with the concept of a ‘universe
of hypotheses’ (i.e. the hypotheses that might be worth testing). One source
which the learner draws on is the LI. Schachter also notes that the universe
expands and contracts during the course of learning, thus allowing for the
utilization of the LI at different stages of development. The hypotheses in a
universe fall into natural groupings, called domains, which Schachter sug¬
gests correspond to abstract linguistic categories such as clause and phrase
types. Domains can vary in size; for example, the general domain of ‘main
verbs’ can be broken down into the narrower domains of ‘main verbs with
complements’ and ‘main verbs without complements’. Learning takes place
when the learner chooses a domain and samples hypotheses within it, testing
them out against the input. The LI, as part of the learner’s existing know¬
ledge, influences both the choice of domain and the specific hypothesis to be
tested; it can contribute to both correct and incorrect hypotheses. Schachter
also acknowledges that constraints, such as the distance between the native
and target languages, affect both how and how much the LI is drawn on.

A framework for explaining first language transfer

Figure 8.3 provides a framework that incorporates both communication and

learning transfer.10 It proposes the following:

1 The LI system is utilized by both comprehension and production mech¬

anisms. In both cases, there are constraints that govern when transfer takes
place.
2 The interlanguage system is also utilized in the processes of comprehending
and receiving messages.
3 The LI system is utilized in the hypothesis construction responsible for
interlanguage development. Again, constraints exist on when transfer
takes place.
 Language transfer 339

4 Comprehensible input, including that input which has been made compre¬
hensible with the help of LI knowledge, serves as a major source of
information for hypothesis construction.
5 L2 output, including that output which has been made comprehensible
with the help of LI knowledge, may be used for hypothesis construction.

The ‘constraints’ referred to in Figure 8.3 are those discussed in the previous
section of this chapter. As we have already noted, the precise ways in which
they delimit the use of the LI in comprehension, production, and hypothesis
construction are not yet fully understood.

L2 L2
input output

Figure 8.3: The role of the LI in L2 communication and learning

Conclusion: problems in the study of transfer

Over the last twenty or so years there have been considerable advances made
in the study of LI transfer, not least in the methods used to investigate it. In
early studies, there was a tendency to claim that any L2 error that showed a
similarity to an LI feature was the result of transfer. Jackson (1981), for in¬
stance, argued that non-inverted wh- questions (such as ‘How I do this?’)
were indicative of the LI influence in Punjabi-speaking learners of English. As
we have seen, such conclusions are not warranted unless it can be shown that
these errors are not ‘developmental’ (i.e. do not occur in the interlanguages of
all learners). There is plenty of evidence that non-inverted wh- questions are
universal features of L2 acquisition. More recent studies have avoided such
misjudgements by carrying out learner comparisons (comparing the output
of learners with Lis that differ with regard to the presence or absence of par¬
ticular linguistic features). Another promising approach, illustrated in Eck-
man (1977) and Gass (1987), involves the bi-directional study of transfer (i.e.
in the case of two languages, x and y, investigating whether a specific feature,
z, is transferred from x to y and also from y to x). Both the crosslinguistic and
the bi-directional approaches not only result in more reliable research but
340 Internal factors

also provide insights not readily available from the more traditional con¬
trastive approach.
There are still a number of problems faced by transfer researchers, how¬
ever. We will consider two: (1) the problem of how to distinguish commun¬
ication and learning transfer, and (2) the problem of how to compare two
languages. Neither problem is specific to the study of transfer, of course. All
L2 acquisition researchers face the general problem of distinguishing what is
strategic—the product of some compensatory strategy—and what is evid¬
ence of the learner’s L2 knowledge system. Also, comparative linguists and
writers of pedagogical grammars have to cope with the theoretical and prac¬
tical difficulties involved in comparing two languages. Both problems, how¬
ever, seem quite central to the work of transfer researchers.
In the last section, we saw that it is possible to make a clear theoretical dis¬
tinction between communication and learning transfer. It is less clear, how¬
ever, whether these two types of transfer can be distinguished empirically,
given that the main data for the study of transfer come from language per¬
formance. The problem is acute because, as Kasper points out, ‘learners
sometimes transfer in want of a better solution even though they consider a
given concept non-transferable’ (1984b: 20). Thus, not all LI features found
in communication transfer will find their way into the learner’s inter¬
language. Not surprisingly, most researchers ignore this problem and assume
that evidence of transfer in performance (usually production) is also evidence
of transfer in learning. One solution lies in collecting introspective data from
the learner, as in Poulisse’s (1990a) study of communication strategies, which
we will consider more fully in the next chapter. In this study, the subjects were
invited to comment retrospectively on the strategies they used to describe
referents for which they had no available L2 word. They frequently com¬
mented on their LI-based strategies without being prompted. Introspective
data may also be helpful in identifying instances of avoidance.
Many of the problems that arise in comparing two languages were evident
in the early days of contrastive analysis. One was that contrastive analyses
generally failed to meet the criteria of descriptive and explanatory adequacy
that any description of language must meet. They faced the same challenges
found in the study of single languages—how to develop comprehensive de¬
scriptions, to what extent the linguistic data should be idealized, and how to
cater for the interaction of linguistic subsystems (the influence that one sys¬
tem, say discourse, has on other systems). There was the problem of equival¬
ence—whether it was possible to find a theoretically sound basis for
comparing two languages. Sajavaara (1981a) listed the procedures that could
be followed: (1) identify linguistic categories common to the two languages
and compare how they are realized in each language, (2) search for equival¬
ents of a given category in one language in the other, (3) compare rules or
hierarchies of rules in the two languages, (4) examine how a given semantic
category is realized in the two languages, and (5) investigate how a given
 Language transfer 341

language function is performed in both languages. All of these procedures

were used, but there was little agreement about which constituted the optimal
method. These problems are perhaps less acute now than earlier, given that
less emphasis is placed on preparing comprehensive contrastive analyses and
more on the detailed examination of specific linguistic elements. However,
even the more focused studies found in current transfer research necessitate
some kind of contrastive analysis, so many of the earlier problems continue to
arise. It is worth noting, though, that transfer studies can also contribute to
linguistic description, Kellerman’s (1989) study of conditionals in the L2
English of advanced Dutch learners being a case in point.11
As well as the difficulty of quantifying transfer, there is the problem of de¬
termining just how significant it is in L2 acquisition. It is now generally ac¬
cepted that although transfer is an important factor in L2 acquisition, it is not
the only factor and often works together with other factors, such as natural
principles of language acquisition. For this reason it may not be inappropriate
to attempt a precise specification of its contribution or to try to compare its
contribution with that of other factors, which was the concern of the early
research in the 1970s. More important, perhaps, is to work towards a fuller
theoretical understanding of how and when learners draw on their LI. The
continued interest in the study of transfer, evident in such publications as
Kellerman and Sharwood Smith (1986), Ringbom (1987), Dechert and
Raupach (1989), and Odlin (1989), testifies to its importance in L2 acquisi¬
tion and has contributed substantially to our understanding of how it
operates.
No theory of L2 acquisition is complete without an account of LI transfer.
This will become apparent in the next two chapters when we consider a vari¬
ety of cognitive theories, such as the Competition Model in which the LI con¬
stitutes a major factor, and also theories based on linguistic universal. In
both, the learner’s LI is a central factor.

Summary

Definitions of transfer
‘Transfer’ is to be seen as a general cover term for a number of different kinds
of influence from languages other than the L2. The study of transfer involves
the study of errors (negative transfer), facilitation (positive transfer), avoid¬
ance of target language forms, and their over-use.

Contrastive analysis and behaviourism

According to behaviourist theories of learning, the LI facilitates learning
where native- and target-language structures are the same, and results in
342 Internal factors

errors where they are different. The strong form of the contrastive analysis
hypothesis (CAH) claims that it is possible to predict when difficulty will oc¬
cur on the basis of the differences between the native and target languages.
Behaviourist views and the strong form of the CAH are no longer considered
tenable on the grounds that many errors are not caused by transfer and that
many predicted errors do not occur. However, contrastive analysis is still an
essential tool in transfer research, particularly if it is supplemented by
comparisons of learners with different language backgrounds.

Minimalist positions
The rejection of the CAH led to the advancement of ‘minimalist positions’ re¬
garding the LI, as reflected in the work of Dulay and Burt, and also to the
Ignorance Hypothesis, according to which the LI functions primarily as a
communication strategy for filling in gaps in the learner’s competence. Min¬
imalist positions underestimate the role of the LI. Even in structural areas
such as basic word order, where it has been claimed transfer is almost non¬
existent, clear evidence for transfer can be found. The ignorance hypothesis is
not tenable given the weight of evidence which suggests that direct learning
transfer does occur.

Current research
Current research is directed at explaining why target-native language
differences sometimes result in transfer and sometimes do not. To this end, a
number of constraints on the transferability of items have been identified; the
language level, sociolinguistic factors of both a macro and micro nature,
linguistic markedness, prototypicality, language distance and the learner’s
psychotypology, and developmental factors involving universal principles or
tendencies in language acquisition. However, although there is clear evidence
that these constraints influence when transfer takes place and also what is
transferred, little is yet known about how the various factors interact.

Transfer in communication
Current theorizing about transfer takes place within a cognitive framework.
It has been suggested that both transfer in communication and transfer in
learning take place and, furthermore, that the former can contribute to the
latter. However, the LI can have a direct effect on interlanguage development
by influencing the hypotheses that learners construct.
Like any other area of SLA research, work on transfer faces a number of
problems, the major one being the difficulty in distinguishing empirically
between instances of communication and learning transfer.
 Language transfer 343

Conclusion

Despite these problems, there is now clear evidence that the LI acts as a major
factor in L2 acquisition. One clear advance in transfer research has been the
reconceptualization of the influence of the LI; whereas in behaviourist ac¬
counts it was seen as an impediment (a cause of errors), in cognitive accounts
it is viewed as a resource which the learner actively draws in interlanguage
development.

Notes

1 U-shaped behaviour involving transfer in an early stage can also be invest¬

igated by means of pseudo-longitudinal studies. These involve the cross-
sectional examination of groups of learners with the same LI but differ¬
ent stages of L2 development. The developmental pattern is evident in the
differences between the groups. One difficulty with this approach, how¬
ever, concerns how to determine what stage of development each learner
in the sample has reached. Given that learners can produce the same cor¬
rect feature at both an early and a late stage of development, this is obvi¬
ously not an easy task.
2 This example of left-branching in Japanese is taken from Kamimoto et al.
(1992:268).
3 Kellerman (1987) also points out that Dulay and Burt were wrong in their
claim that the meanings of ‘interference’ as used in the work of Haugen
and Weinreich on the one hand and in that of the contrastive analysts on
the other were distinct. He provides quotations from Haugen (1956) to
show that, in fact, Haugen’s idea of ‘interference’ was more or less ident¬
ical to that of Lado.
4 Ervin-Tripp’s claim amounts to the assertion that transfer will be more
common in classroom than in naturalistic language learning. This claim is
not supported by research, which in fact suggests the opposite, as is clear
in the discussion of transfer in relation to social setting on page 317.
5 I am grateful to Kellerman for pointing out to me the problems of measur¬
ing the frequency of transfer effects at different language levels.
6 Odlin (1989) also argues that focused learning situations are likely to fa¬
vour positive transfer.
7 I am grateful to Kellerman for helping to clarify the role of social context
in language transfer and, in particular, for pointing out that learners can
behave very differently, where transfer is concerned, inside and outside of
the classroom.
8 It is also possible to distinguish a number of different types of commun¬
ication (production) transfer according to ‘transfer load’; code-shifting
(i.e. the use of an LI word, phrase or sentence in L2 discourse); for-
eignizing (i.e. the use of an LI word in accordance with the L2 sound
344 Internal factors

— LI system

-— code shift
— foreignizing
— literal translation

— interlingual transfer (i.e. learning transfer)

— Interlanguage system

Figure 8.4: Types of production transfer (Kasper 1984b)

system) and literal translation. Kasper (1984b) suggests that these
different types, which seem to apply especially to lexis, can be ordered ac¬
cording to their transfer load as shown in Figure 8.4.
9 Again, I am grateful to Kellerman for pointing out these problems with a
communication-based account of transfer.
10 The model suggests that LI knowledge and interlanguage knowledge are
separate. However, Faerch and Kasper (1989: 178) rightly point out that
‘different linguistic levels may have different types of cognitive repres¬
entation’. Thus, L1/L2 knowledge may have a co-ordinate representation
in the mind of the learner (as is likely for at least part of the lexicon and,
perhaps also for phonology) or a compound representation (likely in the
case of grammar). The type of representation may also be influenced by
the learner’s psychotypology.
11 Kellerman (1989) points out that whereas descriptions of English typic¬
ally insist that ‘would’ does not appear in the subordinate clause of condi¬
tional sentences, there are clear instances of intrusive ‘would’, as for
instance in ‘By the time you’d have noticed it, it’d have been too late’
(from Wong-Fillmore 1985). Thus, the tendency evident in Dutch learn¬
ers’ English interlanguages is also reflected in at least some (generally un¬
acknowledged) LI sentences.

Further reading

The best general account of LI transfer, both because of its extensive cover¬
age of the research and its readability, is T. Odlin, Language Transfer (Cam¬
bridge University Press, 1989).
A balanced and readable account of the contrastive analysis hypothesis can
be found in C. James, Contrastive Analysis (Longman, 1980).
H. Ringbom, The Role of the First Language in Foreign Language Learning
(Multilingual Matters, 1987) provides a good account of the important work
carried out with Swedish- and Finnish-speaking Finns.
There are several useful collections of papers on transfer, the most im¬
portant of which are:
S. Gass and L. Selinker (eds.), Language Transfer in Language Learning
(Newbury House, 1983).
 Language transfer 345

Kellerman, E. and M. Sharwood Smith (eds.), Crosslinguistic Influence in

Second Language Acquisition (Pergamon, 1986).
Dechert, H. and M. Raupach (eds.) Transfer in Production (Ablex, 1989).
The first of these is particularly valuable to readers beginning their study of
transfer, because it includes papers that reflect behaviourist, minimalist, and
cognitivist positions.
Of the many papers that have been published in journals, the following is a
representative sample, reflecting the key issues discussed in this chapter:
F. Eckman, ‘Markedness and the contrastive analysis hypothesis.’ Language
Learning (1977) 27: 315-30.
S. Gass, ‘Language transfer and universal grammatical relations.’ Language
Learning (1979) 29: 327-44.
K. Hyltenstam, ‘Implicational patterns in interlanguage syntax variation.’
Language Learning (1977) 27: 383-411.
E. Kellerman, ‘Transfer and non-transfer: where are we now.’ Studies in Se¬
cond Language Acquisition (1979) 2: 37-57.
J. Schachter, ‘An error in error analysis.’ Language Learning (1974) 24:
205-14.
D. Singleton, ‘Mother and other tongue influence on learner French.’ Studies
in Second Language Acquisition (1987) 9: 327-46.
L. White, ‘Markedness and second language acquisition.’ Studies in Second
Language Acquisition (1987) 9: 261-86.
H. Zobl, ‘Developmental and transfer errors: their common base and (pos¬
sibly) differential effects on subsequent learning.’ TESOL Quarterly (1980)
14: 469-79.
In addition, there is a rich literature dealing with pragmatic transfer (see the
suggestions for further reading at the end of Chapter 5).
■ <
9 Cognitive accounts of second
language acquisition

Introduction

In the last chapter we noted that language transfer has been increasingly un¬
derstood as a cognitive process; that is, L2 learners make strategic use of their
LI in the process of learning the L2, and in the process of understanding and
producing messages in the L2. In this chapter, we will examine a number of
other accounts of L2 acquisition which adopt a broadly ‘cognitivist’ stance in
the sense that they see language acquisition as a mental process involving the
use of strategies that explain how the L2 knowledge system is developed and
used in communication.
As we noted in the Introduction to Part Four, a cognitive theory of lan¬
guage acquisition sees linguistic knowledge as no different in kind from other
types of knowledge, and views the strategies responsible for its development
as general in nature, related to and involved in other kinds of learning. This
perspective contrasts with a linguistic theory of L2 acquisition, which treats
linguistic knowledge as unique and separate from hither knowledge systems,
and acquisition as guided by mechanisms that are (in part at least) specifically
linguistic in nature. It is, of course, not always possible to classify particular
theories of L2 acquisition as exclusively ‘cognitive’ or ‘linguistic’, as often
both perspectives are drawn on. In fact, the two perspectives are not mutually
exclusive, and in all probability, a comprehensive theory of L2 acquisition
will need to incorporate elements from both. It is useful to distinguish the two
perspectives, however, as they reflect clear epistemological differences (see,
for example, the debate between Piaget and Chomsky in Piatelli-Palmarini
1980). This chapter focuses on accounts of L2 acquisition that are broadly
cognitive in nature, while the following chapter examines those that are lin¬
guistic in orientation.
The bulk of this chapter is organized into two major sections, one dealing
with cognitive accounts of L2 learning, and the other with L2 communica¬
tion. In other words, a distinction is made between theories that explain how
learners construct their mental representations of the L2 (i.e. how knowledge
of the rules and items that comprise the L2 is developed) and theories that
explain how learners employ their knowledge in actual language use (i.e.
how L2 comprehension and production is accomplished). This distinction
348 Internal factors

between ‘acquisition’ and ‘procedural ability’ is theoretically clear, but as we

will see, it becomes blurred in theories that view ‘acquisition’ as the concomit¬
ant of ‘use’. Nevertheless, it is a useful tool for surveying the wealth of theo¬
rizing that has taken place. Accordingly, we will focus first on how rules and
items are acquired before examining aspects of L2 procedural ability.
Linguistic knowledge is traditionally described in terms of rules and items,
whether the theory is a cognitive or a linguistic one. However, one school of
cognitive psychology has challenged this assumption, arguing that language
knowledge is best seen as a network, involving a complex set of inter¬
connections between various units (see McClelland, Rumelhart, and the PDP
Research Group 1986). In this theory, therefore, the distinction between
‘learning’ and ‘procedural ability’ no longer seems appropriate, as L2 know¬
ledge and use are accounted for together in terms of ‘connection strength’
rather than separately as rules, items, and processes of language comprehen¬
sion or production. Parallel Distributed Processing, as it is applied to
language acquisition, is examined in the final section of this chapter.
The theoretical positions outlined in this chapter should be seen as af¬
fording multiple perspectives on what are enormously complex phenomena.
The positions often overlap, but not always in ways that can be clearly
specified.

Cognitive accounts of second language acquisition

Linguistic accounts of interlanguage are directed at describing learners’ com¬
petence, conceptualized as an abstract system of rules and items that underlie
actual performance. They are concerned with what learners ‘know’, not with
what they ‘do’. In contrast, although the cognitive accounts to be considered
in this section are still concerned with what the learner ‘knows’, knowledge is
considered to be inseparable from actual use. The focus, therefore, is not on
abstract linguistic knowledge, but on the extent to which the learner has
achieved mastery over the formal and functional properties of language and
the mental processes involved. An assumption of all the theories considered in
this section is that ‘mastery’ is gradable and that there are degrees of ‘know¬
ing’—that learners can, for example, partially know the rule for subject rela-
tivization in English. It is with regard to this notion of ‘mastery’ that the
theories can be seen as cognitive in nature.

A general theoretical framework

A complete synthesis of the various theories and models considered in this

section is not possible, as they reflect widely diverging positions regarding
both which phenomena are in need of an explanation, and how this can best
be provided. However, to give some shape and order to the exposition that
follows, a general framework for investigating L2 acquisition will be outlined
 Cognitive accounts of second language acquisition 349

and an indication given of which parts of this framework the various models
might help us to understand.
The theoretical framework, which is shown diagrammatically in Figure
9.1, is a development of Gass (1988). Gass distinguishes (1) apperceived (or
noticed) input, (2) comprehended input, (3) intake, and (4) integration.
Apperceived input is the first stage of acquisition—a ‘passing through of the
initial data’ (1988: 201). It consists of ‘noticing’ features in the input as a res¬
ult of the saliency of the features themselves and of the learner’s existing L2
knowledge. Not all apperceived input is comprehended (or contributes to the
learner’s understanding of message content). Similarly, not all comprehended
input becomes intake. As Gass puts it, ‘what is comprehended can either feed
into the intake component or, alternatively, it may be not used by the learner
for anything beyond communication’ (1988: 205). Intake, following Chau-
dron (1985), is seen as ‘a process which mediates between target language in¬
put and the learner’s internalized set of rules’ (1988: 206). It does not become
part of the learner’s implicit knowledge system until it has been ‘integrated’.

^ explicit
knowledge
i

▼ y
L2 > noticed —comprehended —intake implicit *-L2
input input input knowledge output
(IL system)

Figure 9.1: A framework for investigating L2 acquisition

These distinctions have been maintained in Figure 9.1. One additional com¬
ponent is proposed—explicit knowledge. In fact, Gass seems to acknowledge
the existence of this when she suggests that some input may be processed and
‘put into storage’ if it is not yet possible to integrate it into the interlanguage
system (1988:207). This ‘storage’, it is suggested, takes the form of some kind
of explicit representation of L2 items and rules. Explicit knowledge, as we
will see later, can contribute to output through monitoring, and also may aid
the processes that contribute to intake. Output can influence input through
interaction, as suggested in Chapter 7.
The different theoretical positions considered in this chapter address differ¬
ent aspects of this framework. Interlanguage theory (Selinker 1972), with
which we will begin, is primarily concerned with implicit L2 knowledge and
the strategies that contribute to its development. Models based on the expli¬
cit/implicit distinction (for example, Krashen’s Monitor Theory), which we
turn to next, are primarily concerned with identifying the relationship be¬
tween these two types of knowledge and how they are used in L2 output.
350 Internal factors

Variability theories (such as those of Tarone 1983 and Ellis 1985c) provide
an account of L2 knowledge in relation to observed L2 output. Variability is
seen as the product of both the differential use of explicit and implicit know¬
ledge, and also the heterogeneous nature of implicit knowledge itself. Variab¬
ility theories that see implicit knowledge in terms of form-function networks
(for example, Ellis 1985c) also shed light on the way learners organize their
implicit knowledge, as do functionalist grammar accounts of L2 learning
(such as those of Klein and Sato). An assumption of both types of theory is
that qualitative as well as quantitative changes take place over time, and that
the introduction of a new item or rule has repercussions for the whole system.
Both the Competition Model (Bates and MacWhinney) and Operating Prin¬
ciples (Andersen) constitute an attempt to specify how input is noticed and in¬
terpreted at different stages of development and how this information is then
organized as implicit knowledge. The Multidimensional Model (Meisel,
Clahsen, and Pienemann) focuses on the relationship between implicit know¬
ledge and output by indicating the strategies which have to be mastered in or¬
der to produce different structures. Finally, skill-learning models (Anderson
and McLaughlin) consider the changes in the way knowledge is represented
as a result of the need to use the L2 efficiently in a range of tasks that place dif¬
ferent demands on the learner’s processing abilities. Some of the models (for
example, the Multidimensional Model) also account for variation in the suc¬
cess of individual learners.
What is common to all the theories and models we will examine in this sec¬
tion is that they attempt to explain the learner’s representation of L2 know¬
ledge and how this changes over time. The models differ in terms of whether
their primary focus is (1) the relationship between input and implicit know¬
ledge, (2) the relationship between explicit and implicit knowledge, (3) the re¬
lationship between L2 knowledge (implicit or explicit) and output, and (4)
whether they try to account for individual learner differences.

Interlanguage theory

The starting point for any discussion of the mental processes responsible for
L2 acquisition is the concept of inter language, a term coined by Selinker
(1972).1 It is used to refer to both the internal system that a learner has con¬
structed at a single point in time (‘an interlanguage’) and to the series of inter¬
connected systems that characterize the learner’s progress over time
(‘interlanguage’ or ‘the interlanguage continuum’). This construct has been
subject to both cognitive and linguistic interpretations, but we will be con¬
cerned with only the former here.
Interlanguage theory is an appropriate starting point because it was the
first major attempt to provide an explanation of L2 acquisition, and many
later theories (such as my and Tarone’s variability models) were develop¬
ments of it. Like all theories, it is dynamic, constantly adapting to new
 Cognitive accounts of second language acquisition 351

information. Early interlanguage theory was informed by the research that

investigated learners’ errors and the general pattern of L2 development (see
Chapters 2 and 3). What follows is a general account of its main premisses,
considered from a cognitive perspective.
The key questions addressed by interlanguage theory are: (1) What pro¬
cesses are responsible for interlanguage construction? (2) What is the nature
of the interlanguage continuum? and (3) What explanation is there for the
fact that most learners do not achieve full target language competence?

Processes of interlanguage construction

In the article that first introduced the term ‘interlanguage’, Selinker (1972)
identifies five principal cognitive processes responsible for L2 acquisition:

1 Language transfer (some, but certainly not all, items, rules and subsys¬
tems of a learner’s interlanguage may be transferred from the first
language).
2 Transfer of training (some interlanguage elements may derive from the
way in which the learners were taught).
3 Strategies of second language learning (Selinker talks about an ‘identifiable
approach by the learner to the material to be learned’, 1972: 37).
4 Strategies of second language communication (‘an identifiable approach
by the learner to communication with native speakers of the TL’, 1972:
_ 37).
5 Overgeneralization of the target language material (some interlanguage
elements are the result of a ‘clear overgeneralization’ of target language
rules and semantic features).

In retrospect, a number of problems with Selinker’s list are apparent. It is not

clear, for instance, why ‘language transfer’ and ‘overgeneralization’ are listed
separately from ‘learning strategies’, as they would both appear to be ex¬
amples of these. However, the list was a valuable one. It constituted one of the
first attempts to specify the mental processes responsible for L2 acquisition,
and also served to introduce a number of key distinctions, such as that be¬
tween ‘learning’ and ‘communication strategies’.
Subsequent work has focused on ‘learning strategies’, defined more
broadly than in Selinker’s initial formulation. Selinker, Swain, and Dumas
(1975), for instance, identify three central strategies: language transfer, over-
generalization, and simplification on the basis of errors found in the speech of
children in a French immersion classroom. These strategies have already re¬
ceived attention; overgeneralization was considered in Chapter 2, simplifica¬
tion in Chapter 3, and transfer in Chapter 8.
At a more general level, interlanguage processes have been discussed in
terms of hypothesis-testing. Corder (1976), for instance, suggested that learn¬
ers form hypotheses about the structural properties of the target language
352 Internal factors

on the basis of the input data they are exposed to. In this way, they build a hy¬
pothetical grammar’ which is then tested receptively and productively.
Hypotheses are confirmed if learners’ interpretations are plausible and their
productions accepted without comment or misunderstanding. They are dis-
confirmed if their understanding is defective and if their output fails to com¬
municate and is corrected. In such cases, learners may restructure their
hypotheses, providing they are sufficiently motivated to do so. One of the
main problems of such accounts is that it is not clear how learners obtain the
linguistic information they need to modify hypotheses during communicative
exchanges (see Chapter 7). As we will see in Chapter 10, linguistic theories of
interlanguage maintain that hypothesis-testing cannot provide an adequate
account of interlanguage development.

The nature of the interlanguage continuum

Cognitive theories of interlanguage postulate that, with the assistance of

learning strategies, learners build mental grammars of the L2. These gram¬
mars account for performance in the same way as a native-speaker grammar;
that is, learners draw on the ‘rules’ they have constructed to interpret and pro¬
duce utterances. Interlanguage is said to be systematic because learners be¬
have ‘grammatically’ in the sense that they draw on the rules they have
internalized—a view that casts doubt on the use of the term ‘error’ itself
(Jakobovits 1970; Cook 1971), as learners’ utterances are only erroneous
with reference to target-language norms, not to the norms of their own
grammars.
These mental grammars are perceived as dynamic and subject to rapid
change. Thus, the interlanguage continuum consists of a series of overlapping
‘grammars’. Each grammar shares some rules with the previously constructed
grammar, but also contains some new or revised rules. A rule has the status of
a ‘hypothesis’. Each grammar or interlanguage is likely to be characterized by
competing rules, or, as Corder (1976) puts it, there will be ‘several concurrent
hypotheses, leading to a set of coexistent approximative systems’. It is this
that accounts for systematic variability in learner performance.
One of the outcomes of this view of the interlanguage continuum is that L2
acquisition is characterized not by ‘simplification’ but by ‘complexification’.
Each grammar the learner builds is more complex than the one that preceded
it. Corder (1977a) has suggested that the learner’s starting point is the same as
in LI acquisition: a ‘basic’ system consisting of lexical items and a few simple
rules for sequencing them. This system constitutes the ‘initial hypothesis’ and
may be universal (i.e. all languages, when stripped down, result in the same
basic system). It follows that L2 acquisition involves a recreation rather than
a restructuring continuum; that is, the starting point is not the full LI which is
gradually replaced by L2 rules and items, but a simple, reduced system of the
LI, which is gradually complexified. Corder suggests that this explains why
 Cognitive accounts of second language acquisition 353

interlanguage systems manifest universal properties, particularly in the early

stages of development (see Chapter 3).
As we will see later, the nature of the learner’s ‘initial hypothesis’ is contro¬
versial. If, as Corder suggests, the starting point is the same as in LI acquisi¬
tion a claim that is in itself controversial—then the question arises as to
whether this starting point is some remembered early version of the L1, which
is complexified through the general process of hypothesis-testing (Corder’s
position and one that is essentially cognitive), or whether it is the innate
knowledge of language which all children bring to the task of learning their
LI, as proposed by Chomsky (1965). In essence, this is an argument between
a cognitivist and a linguistic explanation of L2 acquisition. In addition, as we
saw in Chapter 8, there are grounds for believing that in some aspects of
language (for example, phonology) the starting point may be the LI and that
interlanguage may be, in part at least, a restructuring continuum.

Fossilization

One point on which there is broad agreement is that L2 learners, unlike LI

learners, generally do not reach the same level of competence as native speak¬
ers; their ‘final state’ grammar is not the target-language grammar (see Chap¬
ter 3). Thus, certain rules and items ‘fossilize’. As Selinker (1972) puts it:

Fossilizable linguistic phenomena are linguistic items, rules and subsystems

which speakers of a particular NL will tend to keep in their IL relative to a
particular TL, no matter what the age of the learner or amount of explana¬
tion and instruction he receives in the TL.

Fossilized forms may sometimes seem to disappear but are always likely to re¬
appear in productive language use, a phenomenon known as backsliding.
Selinker talks of an ‘interlanguage norm’ that learners will gravitate towards,
especially when under some kind of pressure. He suggests that there is a psy¬
chological mechanism, which he calls ‘fossilization’, which underlies the pro¬
duction of fossilized items. More commonly, however, the term fossilization
has been used to label the process by which non-target forms become fixed in
interlanguage. 2
A number of studies give support to the prevalence of fossilization. Muk-
katesh (1986), for example, identified a number of persistent errors in the
written production of 80 students at a Jordanian university who had an aver¬
age of eleven years’ instruction in English. Examples of such errors were the
use of simple past instead of simple present, the deletion of ‘be’, and the reten¬
tion of pronominal reflexes in relative clauses. Mukkatesh found that neither
error correction nor explicit grammatical explanation had any effect on these
errors, thus reinforcing the view that certain error types are not susceptible to
de-fossilization. More general evidence of fossilization comes from Higgs and
Clifford (1982), discussed in Chapter 13.
354 Internal factors

A number of possible causes of fossilization have been identified. These are

summarized in Table 9.1. As Selinker and Lamendella (1978) conclude, there
is probably no single cause; both internal and external factors play a role.
Ideally we need to specify the differential contribution of the various factors
and how they interact, but we are a long way from being able to do so.

Factors Description Reference

Internal
1 Age When learners reach a Scovel 1988 (see also
critical age their brains lose Chapter 11).
plasticity, with the result
that certain linguistic
features cannot be
mastered.

2 Lack of desire to As a result of various social Schumann 1978a (see also

acculturate and psychological factors, Chapter 6).
learners make no efforts to
adopt TL cultural norms.

External
1 Communicative pressure
2 Lack of learning
opportunity
3 The nature of the feedback
on learner’s use of L2
r
Persistent pressure to Higgs and Clifford 1982
communicate ideas that
require the use of language
that exceeds the learner’s
linguistic competence
leads to fossilization.
Learners lack opportunities Bickerton 1975
for receiving input and also
for using the L2.
Positive cognitive feedback Vigil and Oiler 1976
(signalling ‘I understand
you’) results in fossilization;
negative feedback
(signalling ‘I don’t
understand you’) helps
avoid fossilization.

Table 9.1: Factors hypothesized to influence fossilization

Subsequent developments in interlanguage theory

Interlanguage theory has helped to shape the development of SLA research by
advancing the notion that learners possess a separate mental grammar that
they draw on in L2 performance. The term ‘interlanguage’ is now used by
theorists of very different persuasions and has become almost theory-neutral.
It can be glossed as the ‘system of implicit L2 knowledge that the learner
develops and systematically amends over time’. The idea of ‘fossilization’ has
 Cognitive accounts of second language acquisition 355

also stuck. However, with the obvious exception of ‘transfer’, Selinker’s ‘stra¬
tegies’—the specifically cognitive dimension of the initial theory—have not
been taken up by theorists. Selinker’s (1992) ‘rediscovery’ of interlanguage
does list ‘training and learning strategies’ and ‘simplification and com-
plexification strategies’ as components of the ‘interlanguage hypothesis’
(1992: 247), but offers little elucidation of them. Instead, as we will see in
Chapter 10, subsequent developments have concentrated on the role of lin¬
guistic universals in interlanguage construction.

Implicit and explicit second language knowledge; the role of

consciousness

Early interlanguage theory did not make a clear distinction between different
types of L2 knowledge, but two other early theories did. Both Krashen’s
Monitor Model and Bialystok’s Theory of L2 Learning owed much to early
interlanguage theory and to the research that it spawned. Krashen and Bialys-
tok were concerned with the role of formal instruction in L2 development and
it was this that led them to distinguish implicit and explicit knowledge. It
should be noted, though, that whereas Bialystok saw her theory as clearly
‘cognitive’ in nature, Krashen drew more directly on linguistic concepts, in
particular Chomsky’s notion of innate linguistic knowledge. However, the
distinction between implicit and explicit knowledge, which is central to
Krashen’s thinking, is a cognitive one. This section will begin with a brief def¬
inition of implicit and explicit knowledge. There follows an account of
Krashen’s and Bialystok’s theories. Finally, the role of consciousness in L2 ac¬
quisition is discussed.

A definition of implicit and explicit knowledge

Explicit knowledge has been defined in different ways, but in SLA research it
is generally used to refer to knowledge that is available to the learner as a con¬
scious representation. It is not the same as ‘metalingual knowledge’ (know¬
ledge of the special terminology for labelling linguistic concepts), although it
is often developed hand in hand with such knowledge. Learners may make
their knowledge explicit either in everyday language or with the help of spe¬
cially learnt ‘technical’ language. They may, for example, be able to explain
the error in sentences like:

* The London is my favourite city.

by saying ‘ “the” is not used with the names of cities’ or they may do so with
the help of grammatical terminology, as in ‘proper nouns like “London” do
not take a definite article’.
There are two types of implicit knowledge, formulaic knowledge and rule-
based knowledge. The former consists of ready-made chunks of language (see
356 Internal factors

Chapter 3). Rule-based implicit knowledge consists of generalized and ab¬

stract structures which have been internalized. In both cases, the knowledge is
intuitive and, therefore, largely hidden; learners are not conscious of what
they know. It becomes manifest only in actual performance.

Krashen’s Monitor Theory

The distinction between these two types of knowledge underlies Krashen’s

Monitor Theory, which is the first and, in many ways, the most central of his
five hypotheses.3 Krashen (1981; 1982) claims that learners possess an ‘ac¬
quired system’ and a ‘learned system’ which are totally separate. The former
is developed by means of acquisition, a subconscious process which arises
when learners are using language for communication. The latter is the result
of learning, the process of paying conscious attention to language in an effort
to understand and memorize rules. It is clear that the acquisition/learning dis¬
tinction mirrors the implicit/explicit distinction, a point that Krashen himself
acknowledges (1982: 10).
Whereas the claim that there are two types of knowledge is not controver¬
sial, Krashen’s insistence that ‘learned’ knowledge is completely separate and
cannot be converted into ‘acquired’ knowledge is. This position has become
known as the non-interface position. Krashen argues that ‘acquired know¬
ledge’ can only be developed when the learner’s attention is focused on mess¬
age conveyance, and that neither practice nor error correction enables
‘learned knowledge’ to become ‘acquired’. Furthermore, he claims that utter¬
ances are initiated by the ‘acquired’ system, and that the ‘learned’ system only
comes into play when learners monitor the output from it. Monitoring is pos¬
sible when learners are focused on form rather than meaning and have suffi¬
cient time to access their ‘learned’ knowledge. Learner's can also modify their
output by means of ‘feel’, using ‘acquired’ knowledge. Krashen has continued
to maintain a non-interface position despite considerable criticism from
McLaughlin (1978b; 1987), Sharwood Smith (1981), and Gregg (1984),
among others. He argues that an interface position does not account for cases
such as ‘P’, an advanced L2 learner who consciously knows rules like 3rd per¬
son -s and yet cannot use them in free speech, nor for other cases of learners
who have ‘acquired’ rules without ever having learned them (Krashen 1985:
39-40).

Bialystok’s view of second language learning

Bialystok’s (1978) theory of L2 learning, which is modelled on Figure 9.2,

does allow for an interface between explicit and implicit knowledge. Accord¬
ing to this theory, implicit knowledge is developed through exposure to com¬
municative language use and is facilitated by the strategy of ‘functional
practising’ (attempts by the learner to maximize exposure to language
 Cognitive accounts of second language acquisition 357

through communication). Explicit knowledge arises when learners focus on

the language code, and is facilitated by ‘formal practising’, which involves
either conscious study of the L2 or attempts to automatize already learnt ex¬
plicit knowledge. In these respects, Bialystok’s theory is the same as Krash-
en’s. It differs, however, in allowing for an interaction between the two types
of knowledge. Formal practising enables explicit knowledge to become impli¬
cit, while inferencing allows explicit knowledge to be derived from implicit.
The model also distinguishes two types of output. Type I output is ‘spontan¬
eous and immediate’, while Type II is ‘deliberate and occurs after a delay’
(Bialystok 1978: 74). As might be expected, Type I relies entirely on implicit
knowledge, whereas Type II involves both implicit and explicit. A feedback
loop from both types allows for continual modification of a response.

Figure 9.2: Model of Second Language Learning (from Bialystok 1978)

Whereas Krashen’s position has remained more or less immutable over the
years, Bialystok’s has undergone considerable revision (see Bialystok 1981a;
1982; 1990a; 1991; and also Hulstijn 1990). The development that concerns
us most here is the reconceptualization of L2 knowledge. In the early model
this was represented as a dichotomy—knowledge was either implicit or expli¬
cit—but in subsequent formulations it is represented in terms of two inter¬
secting continua reflecting the extent to which rules and items are ‘controlled’
or ‘analysed’. Again, Bialystok’s definition of ‘control’ has shifted somewhat.
Whereas initially (for example, in Bialystok 1982), it concerned the ease and
rapidity with which the knowledge can be accessed in differing types of lan¬
guage use, in later formulations (for example, Bialystok and Ryan 1985) it re¬
fers to three different functions: the selection of items of knowledge, their
358 Internal factors

co-ordination, and the extent to which selection and co-ordination can be

carried out automatically.
By ‘analysis’, Bialystok refers to the extent to which the learner has abstrac¬
ted an account of some linguistic phenomenon:

Analysis of knowledge is the process by which mental representations of

this knowledge are built up, structured, and made explicit for the learner.
(Bialystok 1991: 65)

One way in which this can take place is by analysing formulas (i.e. dis¬
covering the parts that make them up). It is tempting to see this ‘analysis’ di¬
mension as equivalent to the explicit/implicit distinction, with analysed
knowledge corresponding to explicit knowledge and unanalysed to implicit.
Bialystok, in fact, does equate analysis with the development of an explicit
representation of knowledge, but she emphasizes that analysed knowledge
need not involve consciousness. As she puts it ‘a criterion of consciousness
seriously underestimates the level of analysis with which linguistic knowledge
is represented (1991: 68).
In explaining how analysed knowledge is developed, Bialystok (1991)
draws on the work of Karmiloff-Smith’s (1986) three phases of skill develop¬
ment. The first phase is called ‘Implicit’. Here knowledge of a linguistic item is
closely associated with procedures for using it in communication and is not
represented independently. For example, the learner may supply linguistic de¬
terminers with a variety of nouns and yet not have organized them into a sys¬
tem of determiners. The second phase is called ‘Explicit 1’, during which
learners examine, analyse, and organize their performances in order to con¬
struct explicit and independent representations of linguistic knowledge. In
the final phase—Explicit 2—linguistic knowledge is available for conscious
consideration. This model was intended to account for language develop¬
ment in children, and thus progress is reflected in movement from Implicit to
Explicit 1 and finally to Explicit 2. It is not clear how—or whether—this
model can be applied to L2 acquisition, nor does Bialystok consider this in her
later work.
The goal of much of Bialystok’s later work is to show the relationship be¬
tween different types of knowledge and different types of language use (con¬
versations, tests, reading, studying, etc.). Hulstijn (1990: 38), in fact,
characterizes it as ‘a functional model, aimed at explaining language use in
terms of task demands’ rather than a theory of acquisition. The primary effect
of analysing knowledge is to increase the potential for use in cognitively de¬
manding tasks. Unanalysed knowledge also has its use—in conversation, for
example. This aspect of Bialystok’s research mirrors Cummins’ work on lan¬
guage proficiency (see Chapter 6).
There are a number of problems with Bialystok’s views of language
acquisition (see Hulstijn 1990). In particular, the claim that language must
begin with unanalysed knowledge seems unwarranted in the case of L2
 Cognitive accounts of second language acquisition 359

acquisition. Many instructed L2 learners begin with explicit knowledge. Hul-

stijn comes out strongly in favour of the kind of skill-learning theory of L2 ac¬
quisition that we will consider later (see page 388).

The legitimacy of the explicit/implicit distinction

Not all L2 acquisition researchers have been happy with the implicit/explicit
distinction. McLaughlin (1987: 21) has argued that Krashen’s acquired/
learnt distinction is not tenable because it cannot be falsified; Krashen has
failed to provide adequate definitions of what he means by ‘subconscious’
and ‘conscious’, and ‘he has provided no way of independently determining
whether a given process involves acquisition or learning’. McLaughlin’s criti¬
cisms, however, appear to be levelled primarily at Krashen’s attempt to dis¬
tinguish ‘acquired’ and ‘learnt’ knowledge at the level of process, but as
Bialystok (1981a) has noted, the existence of two types of knowledge is
widely recognized in cognitive psychology. The distinction is not especially
problematic at the level of product, therefore. Furthermore, there have been a
number of successful attempts to operationalize the distinction between ‘im¬
plicit’ and ‘explicit’ learning (for example, Reber 1976).
Examples of studies that have sought to examine learners’ implicit and ex¬
plicit knowledge of grammatical rules are Seliger (1979), Tucker, Lambert
and Rigault (1977), Hulstijn and Hulstijn (1984), Sorace (1985) and Green
and Hecht (1992). In all of these studies explicit knowledge was equated with
learners’ oral or written explanations of grammatical rules, while implicit
knowledge was determined by examining the learners’ use of the same gram¬
matical features in some kind of performance. The main question the studies
sought to answer was ‘What is the relationship, if any, between learners’ ex¬
plicit and implicit knowledge?’ We will focus on the Green and Hecht (1992)
study here, as it provides a comprehensive examination of this question.
The subjects were 300 German learners of English with between three and
twelve years’ exposure to formal teaching, and, also, 50 native English speak¬
ers. The subjects were shown sentences containing different kinds of gram¬
matical errors and asked to correct each sentence and in each case to state the
rule that had been violated. The native speakers were able to correct 96 per
cent of the errors, while the German learners corrected 78 per cent overall.
The most advanced group of learners, however, corrected 97 per cent. Green
and Hecht found that the German learners could only state the correct expli¬
cit rule in less than half the cases (46 per cent). Again, the more experienced
learners showed higher levels of explicit knowledge, the university students
with the most experience formulating correct rules in 85 per cent of the cases.
Although the learners nearly always produced an accurate correction when
they had produced a correct rule (97 per cent of such cases), they were able to
make successful corrections without recourse to explicit knowledge in 43 per
cent of the cases. This study suggests that the learners relied primarily on
360 Internal factors

implicit knowledge, but that the availability of accurate explicit rules facilit¬
ated performance of the correction task. Explicit rules obviously constitute
only a subset—and a fairly small one at that—of available implicit know¬
ledge, a point which Krashen has emphasized (see Krashen 1982).
Whereas the Green and Hecht study examined the implicit/explicit distinc¬
tion at the level of product (the way knowledge is represented in the mind of
the learner), other studies have tried to investigate it in terms of process (how
the two types of knowledge are internalized). N. Ellis (1991), for instance,
draws on earlier studies in cognitive psychology (for example, Reber 1976;
Reber, Kassin, Lewis, and Cantor 1980) as well as SLA theorizing. Interest¬
ingly, Ellis produces definitions of the two types of learning that are very sim¬
ilar to Krashen’s definitions of acquisition and learning:

Implicit learning is coming to learn the underlying structure of a complex

stimulus environment by a process which takes place naturally, simply and
without conscious operations. Explicit learning is a more conscious opera¬
tion where the individual makes and tests hypotheses.

The results obtained by this and other studies are considered in some detail in
Chapter 14.
There would seem to be a reasonable basis in both theory and research for
the distinction at both the level of product and process. In this respect, there¬
fore, Krashen’s acquisition/learning hypothesis is vindicated. It should be
noted, however, that in a number of respects Krashen appears to have been
mistaken. First, there is evidence from the Green and Hecht study that learn¬
ers are capable of learning a substantial number of explicit rules. Krashen’s
(1982: 98ff.) claim, therefore, that ‘learning’ is limited to a few ‘simple’ rules
is not warranted. Second, there is evidence that ‘learning’ through an explicit
presentation of a rule can sometimes work more effectively than ‘acquiring’ a
rule implicitly.
Accepting the acquisition/learning distinction, however, does not entail ac¬
cepting the non-interface position. Krashen bases this on descriptive work in
SLA research which shows that learners follow a ‘natural order’ of develop¬
ment (see Chapter 3) and some early studies of the effects of formal instruc¬
tion which indicate that instructed learners follow the same order and make
the same kinds of errors, irrespective of the instruction they receive (see Tur¬
ner 1979; Felix 1981; Pica 1983). However, as we saw in Chapter 3, the
whole notion of ‘natural order’ is suspect. Also, more recent studies (for ex¬
ample, Pienemann 1989; White, Spada, Lightbown, and Ranta 1991) have
shown that learners, with certain constraints, can learn what they are taught.
Again, the evidence for these claims is considered in detail in Chapter 14. The
important point here is to note that these studies suggest that ‘learning’ can
sometimes turn into ‘acquisition’.
What then is the nature of the relationship between explicit and implicit
knowledge? Does explicit knowledge convert into implicit knowledge
 Cognitive accounts of second language acquisition 361

through practice, as claimed by Bialystok and argued by Sharwood Smith

(1981), or does it, in the main, only facilitate the acquisition of implicit know¬
ledge, as suggested by Terrell (1991), and myself in Ellis 1993a? These two al¬
ternatives are considered in Chapter 14 under the headings the interface
hypothesis and the selective attention hypothesis.

‘Consciousness’ and ‘noticing’

Underlying the whole question of the relationship between explicit and impli¬
cit knowledge and how they are internalized is the question of ‘consciousness’
in language learning. As Schmidt (1990) points out, this is a controversial
issue. He distinguishes three senses of ‘conscious’. First there is ‘conscious¬
ness as awareness’. Here there are degrees or levels of awareness. ‘Perception’
is not necessarily conscious, but ‘noticing’, which Schmidt defines as ‘availab¬
ility for verbal report’, requires focal awareness, while ‘understanding’ in¬
volves conscious analysis and comparison with what has been noticed on
previous occasions. Second, there is ‘consciousness as intention’. Schmidt ac¬
cepts that not all intentions are conscious. Finally, there is ‘consciousness as
knowledge’. Schmidt suggests that the explicit/implicit contrast represents a
continuum, but notes that there is no consensus on where to draw the line to
demarcate conscious knowledge.
The general position that Schmidt adopts is that the role of unconscious
learning has been exaggeratedTHeemphasizes that ‘subliminal language learn¬
ing’ is impossible and that some degree of consciousness is necessary for ‘noti¬
cing’ to take place. Schmidt and Frota (1986) examined one learner’s (R’s)
diary to establish which features in the input R had consciously attended to.
They also examined R’s output to see to what extent the noticed forms turned
up in communicative speech. In nearly every case the forms that R produced
were those that he noticed people saying to him. Conversely, forms that were
present in comprehensible input did not show up until they had been noticed.
Noticing is of considerable theoretical importance because it accounts for
which features in the input are attended to and so become intake (information
stored in temporary memory which may or may not be subsequently accom¬
modated in the interlanguage system). Schmidt and Frota suggest that for no¬
ticed input to become intake, learners have to carry out a comparison of what
they have observed in the input and what they themselves are typically pro¬
ducing on the basis of their current interlanguage system. They refer to this as
‘noticing the gap’, and argue that this too is a conscious process.
Other research relevant to the role of ‘noticing’ is Hulstijn’s (1989b) study
of implicit and incidental F2 learning. In two studies involving a natural lan¬
guage (F2 Dutch) and an artificial language, learners were presented with
word order structures implicitly (i.e. the structures were not explained to
them) and incidentally (i.e. they did not know they would be tested for recall
of the structures). They were assigned to one of three treatments involving
362 Internal factors

exposure to sentences containing the target structures. One group (the form-
focused group) had to perform an anagram task that directed their attention
to the structure without any need to consider its meaning. The second group
(the meaning-focused group) were shown the same sentences on a screen and
asked to respond meaningfully to them by saying ‘yes’, ‘perhaps’ or ‘I don’t
know’. The third group (the form- and meaning-focused group) were simply
told to pay attention to both form and meaning but were given no special task
to perform. The results showed that the form-focused group outperformed
the other two groups in terms of gains in scores on a sentence-copying task
and a task requiring cued recall of the sentences used in the learning tasks.
Hulstijn interprets the results as showing that attention to form when encod¬
ing input is a ‘sufficient condition’ for implicit and incidental learning. How¬
ever, as the meaning-focused group also produced significant gains, the
hypothesis that exclusive attention to meaning will inhibit acquisition was
not supported. It is possible, though, that the learners in this group engaged in
some degree of ‘noticing’ of the target structures. As Hulstijn points out,
meaning may be the learner’s first priority, but attention to form occurs as a
‘backup procedure’ in case meaning fails to provide an adequate
interpretation.
This discussion of consciousness in L2 acquisition suggests that the distinc¬
tion between conscious ‘learning’ and subconscious ‘acquisition’ is overly
simplistic. It is clear that ‘acquisition’, in the sense intended by Krashen, can
involve at least some degree of consciousness (in noticing and noticing the
gap). But it is also clear, if the evidence from studies of implicit learning is to
be accepted, that learning can take place without learners being aware of it.4
At the level of product the explicit/implicit distinction seems less problematic.
Clearly, learners may know a rule, or know about it, or both.
The relationship between explicit and implicit knowledge, then, continues
to be a key issue. One possibility, suggested by both Schmidt and Frota (1986)
and myself (Ellis 1993a), is that explicit knowledge functions as a facilitator,
helping learners to notice features in the input which they would otherwise
miss and also to compare what they notice with what they produce. In a sense,
then, explicit knowledge may contribute to ‘intake enhancement’, but it will
only be one of several factors that does this (see Chapter 7 for a fuller discus¬
sion of this point).

Summary

The explicit/implicit distinction is controversial, as it depends on the idea of

‘consciousness’, which for some researchers remains too elusive a concept to
be criterial (see McLaughlin 1990c). The following are the main claims that
have been made:

1 Learners possess two kinds of knowledge, explicit and implicit. This claim
is widely accepted.
 Cognitive accounts of second language acquisition 363

2 Learners can internalize L2 knowledge both explicitly (Krashen’s ‘learn¬

ing’) and implicitly (‘acquisition’). This claim is less widely accepted, al¬
though a number of studies suggest that the distinction between implicit
and explicit is valid at the level of process (for example, N. Ellis 1991)
3 An alternative to the explicit/implicit distinction is to view knowledge as
more or less ‘analysed’ and more or less ‘controlled’ (as in Bialystok’s view
of language learning).
4 Neither a non-interface nor a strong interface model satisfactorily ac¬
counts for the relationship between explicit and implicit knowledge.
5 Increasingly, explicit knowledge is being viewed as a facilitator of implicit
knowledge, by enabling learners to notice features in the input and
compare them with their own interlanguage representations (Schmidt
1990)
6 Different types of language use typically require different kinds of know¬
ledge (Bialystok 1982).

One possible way of resolving some of the problems of the explicit/implicit

distinction can be found in the information-processing accounts of L2 ac¬
quisition considered in a later section of this chapter (see page 389).

Variability theories of second language acquisition

The theories based on the implicit/explicit distinction provided a means for
explaining the variability that many researchers observed in learner-
language: learners varied because they sometimes drew on their implicit
knowledge and sometimes on their explicit. A number of other researchers
(such as Tarone and myself), who were also influenced by early interlanguage
theory, were unconvinced by this explanation and developed alternative the¬
ories to account for variability. Although these theories drew extensively on
concepts from sociolinguistics, they were essentially cognitive in nature as
they sought to account for the way in which learners organized their L2
knowledge and the strategies they deployed for both learning and using it.
Whereas in Chapter 4 we concentrated on identifying different types of
variability and the factors that have been found to induce variability in in¬
terlanguage use, here we address the question ‘How is variability related to
L2 acquisition?’ We will examine my and Tarone’s theoretical positions, and
also consider Preston’s more explicitly sociolinguistic theory.

Tarone’s ‘Capability Continuum’

Tarone (1983) considers three paradigms for studying interlanguage.
According to the ‘Homogeneous Competence Paradigm’, as reflected in
Adjemian (1976):
learners have grammatical intuitions which the linguist may use as data in
modelling that competence. Variation is a phenomenon which occurs in
364 Internal factors

speech performance and not in the grammatical intuitions on the basis of

which the ‘grammar itself’ is written (1976:150).

This paradigm, which is the one that informs the linguistic theories to be con¬
sidered in the following chapter, is inadequate according to Tarone, because
it does not satisfactorily account for the results of variability research, which
show that the careful style is more permeable to invasion from the target lan¬
guage than the vernacular style. The Homogeneous Competence Paradigm
predicts the opposite.
Krashen’s Monitor Model is illustrative of the ‘Dual Knowledge Para¬
digm’, according to which there are two knowledge systems, both of which
are homogeneous. Variability is explained by ‘monitoring’, the process by
which learners modify utterances generated by ‘acquired’ knowledge. There
is no recognition of variability within ‘acquired’ knowledge itself.5 Tarone’s
main criticism of this paradigm is that the research findings show more than
just two dichotomous styles in learner language. Thus the Monitor Model
does not account for the ‘inherent variability in the system on any observed
occasion’ (1983:159).
Tarone’s own model, the Capability Continuum, is based on the Labovian
paradigm, which was described in some detail in Chapter 4. ‘Capability’ is
preferred to ‘competence’ because she needs a term that refers broadly to the
linguistic knowledge that underlies ‘all regular language behaviour’ (1983:
151). The learner’s capability, therefore, is evident in the regularities ob¬
served in production and perception, writing and reading, and making judge¬
ments on grammaticality. Tarone suggests that it is composed of ‘regularities’
(defined as ‘patterns which underlie phenomena in observed behaviour’)
rather than ‘rules’ (defined as ‘normative standards of behaviour’), but still
constitutes ‘an abstract linguistic system’ which exists apart from its use
(1983: 151-2). Capability consists of a continuum of styles, ranged from the
‘careful’ to the ‘vernacular’, and is, therefore, heterogeneous (see Figure 4.2).
The vernacular style is considered to be ‘primary’ in the sense that it is the
most stable and consistent.
Tarone’s theory is more than just an attempt to model the kind of L2
knowledge that learners internalize. It also provides an explanation of how
knowledge is acquired and, importantly for any theory of L2 acquisition,
how changes in the learner’s interlanguage take place. It posits that new
forms enter interlanguage in two ways: (1) directly into the learner’s vernacu¬
lar style, in which case they may subsequently ‘spread’ to more formal styles
over time, and (2) initially into the learner’s most formal style, manifest only
when the learner is paying close attention to speech production, and sub¬
sequently by ‘spreading’ into the less formal styles where they replace those
forms that entered these styles earlier. In the case of (1), there may be a tend¬
ency for the new forms to appear in a ‘universal order’.
 Cognitive accounts of second language acquisition 365

Whereas Tarone’s 1983 paper does not offer much detail about the key no¬
tion of ‘spreading’, her 1988 book does, drawing on the work of Dickerson
(1975) and Gatbonton (1978). Dickerson’s work suggests that learners initi¬
ate change in one linguistic environment, which then spreads to other lin¬
guistic environments in a clear order. Thus, learners move systematically
towards target-language norms over time. Whereas Dickerson’s view of
‘spreading’ allows for variation at the beginning of the acquisition process,
Gatbonton’s sees acquisition as proceeding in two phases: in the acquisition
phase and the replacement phase. Tarone does not offer any views as to which
of these interpretations of ‘spreading’ is the preferred one.

Ellis’ Variable Competence Model

In my account of L2 variability, I have assumed that the way language is
learnt is a reflection of the way it is used. There are two strands to this posi¬
tion. One draws on the idea of a relationship between a differentiated know¬
ledge store and different types of language use, and the other on the idea of
form-function networks.
Like Bialystok, I see L2 knowledge as represented differently in the mind of
the learner according to how analysed and how automatic it is (Ellis 1984a).
The unanalysed/analysed and the non-automatic/automatic distinctions con¬
stitute intersecting continua in which any L2 item or rule is located. In the
case of native-speaker competence, the most usual form of representation is
unanalysed/automatic, although as the discussion of Bialystok’s model in the
previous section made clear, native speakers also have access to analysed
knowledge. Language use is differentiated according to the amount of plan¬
ning that takes place. Thus, following Ochs (1979), it can be planned or
unplanned. Planned discourse is discourse that is thought out prior to expres¬
sion, whereas unplanned discourse lacks forethought and preparation.
Unplanned discourse can be considered primary in that it is the type found in
everyday communication and spontaneous conversation.
Knowledge is activated for use by means of ‘processes’, which, again, are
distinguished according to whether they are primary or secondary. Primary
processes are utilized when learners wish to engage in unplanned discourse.
They draw on knowledge that is relatively unanalysed and automatic—se¬
mantic simplification (see Chapter 3 and Ellis 1984a) is an example. Second¬
ary processes are utilized in planned discourse and draw on knowledge
towards the analysed end of the continuum—monitoring (the conscious edit¬
ing of language performance) is an example.
I also suggest that the processes that the learner calls on in order to particip¬
ate in discourse are developmental, that is, their prominence in L2 acquisition
coincides with the general stages of development discussed in Chapter 3.
Thus, early on, learners make heavy use of semantic simplification because
this process requires little L2 knowledge. Also, L2 knowledge that to begin
366 Internal factors

with can only be used via secondary processes because it exists only in ana¬
lysed form, can eventually be accessed through primary processes and so be¬
comes available for use in unplanned as well as planned discourse.
According to this Variable Competence Model, L2 development takes two
forms. Learners learn how to activate items and rules that are available ini¬
tially only in planned discourse for use in unplanned discourse. In this respect,
my position is similar to that of Tarone. Also, learners acquire new L2 rules
through participating in different types of discourse. In other words, as Wid-
dowson (1979b: 62) has suggested, ‘we create discourse and commonly bring
new rules into existence by so doing’. It is this latter type of development that I
explored in my subsequent work on variability by drawing on the idea of
form-function networks.
The starting point for my ideas on how learners construct variable in¬
terlanguage systems as a result of participating in discourse is free variation
(see Chapter 5 for an account of this). In Ellis 1985c, I suggest that the learn¬
er’s interlanguage is composed of competing rules at any stage of its develop¬
ment. In some cases, these competing rules are systematic, as they relate to
situational and contextual factors. In other cases, the competing forms are
used arbitrarily, in free variation. I argue that new linguistic forms emerge in
all natural languages ‘quite spontaneously’, and that interlanguage is no dif¬
ferent. However, it is inefficient to operate a system in which two forms have
total identity of function, so learners seek to remove free variation by (1) elim¬
inating forms that are deemed non-standard or unnecessary, and (2) building
form-function networks in which different forms are used to perform differ¬
ent functions. Like Tarone, I draw on Gatbonton’s (1978) ideas of‘diffusion’
to explain how form-function networks are constructed. L2 acquisition in¬
volves a first stage (the ‘acquisition phase’), where new forms are acquired
and used in free variation, and subsequent stages (the ‘reorganization phase’)
where learners sort these forms into functional pigeon-holes. The initial
form-function correlations that learners establish are not likely to correspond
to those found in the target language. This requires several sortings and may
never be achieved.
I have not fully drawn together the two strands of my theorizing relating to
(1) the knowledge types and processes involved in different kinds of dis¬
course, and (2) the construction of form-function networks about variability.
Whereas my earlier work drew on psycholinguistic models of L2 acquisition,
the later work was strongly influenced by Huebner’s ‘dynamic paradigm’
(Huebner 1979 and 1983; see Chapter 4) and belongs to a functional account
of interlanguage.

Preston’s Sociolinguistic Model

Both Tarone’s and my models are based on the psycholinguistic notions of

‘attention’ and ‘planning’. However, there is a clear need to incorporate an
 Cognitive accounts of second language acquisition 367

interactional/social dimension to account fully for interlanguage variability.

It was this that led me to examine form-function networks. Preston’s Socio-
linguistic Model includes both dimensions and constitutes the most compre¬
hensive variability theory to date.
Preston (1989) envisages interlanguage development, like other kinds of
language change, as involving both ‘change from above’ and ‘change from be¬
low’. In the case of general language change, the former ‘involves linguistic re¬
sponses to straightforward social pressures’ and is accompanied by
awareness of features that are prestigious or stigmatized, while the latter in¬
volves changes that are not conscious and which usually arise spontaneously
in the speech of members of the working class. In the case of interlanguage,
the vernacular style can be influenced both from above (when forms enter
through the learner’s careful style) and from below (when they enter directly
into the vernacular style). Forms that enter from above are ‘odd’ and ‘require
more effort and attention to maintain’. Examples are English third person
singular -s and noun plurals. The distinction between these two types of
change is necessary to account for the finding that the direction of stylistic
shifting varies, with some features (such as articles) showing greater accuracy
in the vernacular and others (like third person -s) greater accuracy in the care¬
ful style (see Tarone 1985 and Chapter 4).
Preston’s model, which is shown in Figure 9.3, rests on the idea of the
learner’s knowledge of the L2 as ‘a complex variation space’ (1989: 265),
which can be accounted for in terms of (1) planning, (2) depth, and (3) stabil¬
ity. The concept of planning is envisaged as a continuum, as in my model.
Preston, however, sees ‘planning’ not just as something that learners do when
using their knowledge, but as actually reflected in the knowledge system it¬
self. He talks of the ‘planned and unplanned sides of learner systems’ in a dis¬
tinction that seems very close to Bialystok’s ideas of ‘analysed’ and
‘unanalysed’ knowledge. Learners will vary as to which side—or type of
knowledge—is most fully represented in their interlanguage systems.
The ‘depth’ dimension of the model is an attempt to take account of the so¬
cial uses that learners make of their variable systems. Thus, there are likely to
be different frequencies of use that reflect gender, class, age, genres, relation¬
ships, etc. These differences reflect the attempts of learners to use their know¬
ledge functionally for social purposes. Learners, like native speakers, will use
forms symbolically for these purposes.
‘Stability’ is a characteristic of both the surface structure of the learner’s in¬
terlanguage system and also of the way this system is used to convey social
meanings. In the case of the former, a stable system is one where there is an
absence of variation because no new forms are entering the system and ex¬
isting forms have become categorical. In the case of the latter, ‘stability’ is
evident when continued association of a given feature with a given social
meaning halts or slows the development of that feature. When this kind of
stability is found, fossilization may occur.
368 Internal factors

+ ■<-Planning-► -

Figure 9.3: An integrated model of language variation (from Preston 1989)

Preston raises the interesting possibility that learners will differ in the ex¬
tent to which the variability of their systems is a product of the planning or the
depth dimensions:

Where there is no real community of learners and no previous accretion of

symbolic depth,... change will be more linguistically determined along the
shallow, surface plane of the model proposed here.

Thus, classroom learners are more likely to manifest planning than social
variability, although, as Preston is careful to point out, even these learners are
likely to assign some symbolic value to developing forms.

Evaluation
This review of three variability theories demonstrates the need for models of
considerable complexity if all the known sources of variability are to be ac¬
counted for. It is probably true to say that, as yet, there is no complete theory
of interlanguage variation. Such a theory will need to explain (1) the cognitive
processes involved in planning variability, (2) the nature of the form-function
networks that learners construct at different stages of development, and (3)
the systematic way in which learners use L2 knowledge to convey social
meanings. Tarone’s and my own theories are concerned almost entirely with
(1), and hence their inclusion in a chapter dealing with cognitive accounts of
L2 acquisition. My later work also considered (2). Preston’s theory is the
most comprehensive in that it accounts for both (1) and (3) and constitutes,
 Cognitive accounts of second language acquisition 369

therefore, a real attempt to integrate psycholinguistic and sociolinguistic per¬

spectives. However, as presented in Preston (1989), it is not always very clear
and does not appear to provide an explanation of (2). All three theories offer
only partial explanations, therefore, although in this respect they probably do
not differ from any other interlanguage theory.

Functionalist theories of second language acquisition

Functionalist theories of L2 acquisition share a number of concerns with vari¬

ability theories. For instance, both are concerned not just with how linguistic
knowledge is represented in the mind of the learner, but also with how this
knowledge is used in discourse. Also, both types assume that syntax cannot
be considered separately from semantics and pragmatics and, as such, are op¬
posed to purely linguistic accounts of L2 acquisition.
According to Tomlin (1990):

A general premise of FAs (functional approaches) is that the acquisition of

a language arises from general circumstances of use and communicative in¬
teraction (1990: 161).

He distinguishes strong and weak functional views. In the former, language

acquisition is determined solely by general learning mechanisms operating on
the rich data provided by human interaction. In the latter, some linguistic
knowledge is considered to be available from the outset, but the way this is
manifested in specific languages must be discovered through participation in
discourse.
In this section, we will consider a number of functional theories of L2 ac¬
quisition. We will begin examining the functional approach adopted by a
number of researchers involved in the European Science Foundation Project
in Second Language Acquisition by Adult Immigrants (for example, Perdue
1984 and 1991; Klein and Perdue 1989; Perdue and Klein 1992). We will
then examine work informed by Givon’s Functional-Typological Theory (for
example, Sato 1988; Pfaff 1992). In both cases, the testing of the theories has
led to detailed and painstaking description of L2 learners’ productions in nat¬
ural discourse, undertaken with a view to discovering the organizational
principles that operate in learner-language. Thirdly, we will consider Bates
and MacWhinney’s Competition Model (for example, Bates and Mac-
Whinney 1982 and 1987). This theory has afforded some very specific
hypotheses, which have been tested experimentally.

The learner’s four tasks

Klein (1986) characterizes the learner’s task as that of overcoming four prob¬
lems. These are described in Table 9.2. The learner faces these problems at
one and the same time and must cope with them together. Thus, the analysis,
370 Internal factors

synthetic, embedding, and matching problems are all interrelated and in¬
terdependent. Learners solve them by utilizing shared world knowledge, situ¬
ational knowledge, and contextual information from the preceding discourse
to help them understand input and produce output and, by so doing, they de¬
velop their interlanguages.
Learning a language involves mastering a number of fundamental func¬
tions of language—spatial and temporal reference, for example. The LI
learner has to learn both the functions themselves and the means for per¬
forming them. But in L2 acquisition, the learner already knows these
functions and brings them to the input. Therefore, according to Klein (1991):
It is these functions ... which drive the learner to break down parts of the
input and to organize them into small subsystems, which is reorganized
whenever a new piece from the flood of input is added, until eventually the
target system is reached (or more or less approximated) (1991: 220).
In other words, the solution to the four problems lies in the learner’s know¬
ledge of functional meanings. For Klein, L2 acquisition is primarily function¬
ally driven.

Problem Description

Analysis The learner has ‘to segment the stream of acoustic signals into
constituent units and to bring the latter into line with the parallel
information on concurrent events which constitutes the situational
context of the utterance’ (1986: 59).
Synthesis The learner has to try to put the sounds and words he has learnt together
in order to produce and comprehend L2 utterances.
Embedding The learner has to make utterances fit the context—situational and
linguistic—in which they occur. This requires ‘a sort of balance of
linguistic and contextual information’ (1986: 61).
Matching ‘... the learner must continuously compare his current language variety
with the target variety’ (1986: 62).
I

Table 9.2: The learner’s four tasks in L2 acquisition (based on Klein 1986: 59-62)

A number of studies have demonstrated how L2 learners handle functions

like temporal reference in the early stages. Klein (1986), for instance, de¬
scribes the devices used by an Italian woman with very limited knowledge of
German in telling a story about how her husband was involved in an indus¬
trial accident. The learner manifested almost no knowledge of German verb
tenses and yet was able to convey successfully temporality by using tech¬
niques that relied on a small set of linguistic devices (for example, simple ad-
verbials like ‘dann’ and ‘morgen’) and shared world knowledge. Perdue
(1991) reports more generally on how the subjects studied in the European
Science Foundation Project on Adult Second Language Acquisition handled
spatial and temporal reference. The learners acquired a few simple words to
 Cognitive accounts of second language acquisition 371

express ‘essential’ reference and relations (for example, ‘up/above’ and ‘left/-
right’), they used transparent form-meaning relationships, they decomposed
complex relations into simpler ones (for example, instead of ‘between two
chairs’ they used circumlocutions like ‘side of chair, side of other chair,
middle’), and they relied on the inferencing capacities of their interlocutors.
Perdue considers that learners who manifest such tendencies have access to
only‘a “basic” learner language’ (1991: 418), which they progress beyond in
order to escape from ‘the expressive constraints of a simple system’ and ‘to
match the target language more closely’ (1991: 419).
This idea of a ‘basic variety’ is explored further by Perdue and Klein (1992),
in a study of two adult Italian learners of L2 English in London. Both learners
began by producing utterances which were very simple, characterized by the
use of strings of simple noun phrases and a limited number of adjectives and
adverbials. They relied heavily on interlocutor scaffolding. Their utterances
contained no case marking, few verbs, which were almost always in the base
form, and hardly any copulas. One of the learners, Andrea, proceeded to
‘grammaticalize’ his speech over a period of about 20 months, whereas the
other, Santo, maintained the ‘basic variety’ throughout. Grammaticalization
took the form of the abandonment of a commonly used non-standard struc¬
ture V + NP (as in ‘have the one family’ = ‘there is/was a family) and the devel¬
opment of systematic verb morphology, case-markings for third-person
pronouns, and target-like subordinate constructions. Perdue and Klein ex¬
plain the differences in the development of the two learners in terms of the
‘communicative style’ they typically adopted in interactions with native
speakers. Thus, whereas Santo tended to take charge of these interactions,
Andrea generally avoided initiating topics, but was ‘very receptive to native
speakers’ reformulations of his utterances’ (1992: 270).

Language acquisition as syntactization

Like Klein, Givon (1979) also sees syntax as inextricably linked to dis¬
course—it is ‘a dependent, functionally motivated entity’ in the sense that its
formal properties reflect its communicative uses. He distinguishes two types
of language, each with its own type of structures. There are the loose, paratac-
tic structures found in informal/unplanned discourse, which constitute the
‘pragmatic mode’ (later referred to as the ‘pre-grammatical mode’ in Givon
1995). There are also the tight, ‘grammaticalized’ structures found in
formal/planned discourse, which constitute the ‘syntactic mode’ (later called
the ‘grammatical mode’). An example of the former is the topic-comment
structure of an utterance like:
Ice cream, I like it.
while an example of the latter is the subject-predicate structure of an utter¬
ance like:
I like ice cream.
372 Internal factors

Table 9.3 lists the main differences between the two types of language use.
Givon argues that acquisition is characterized by syntactization, i.e. the
gradual move from a pre-grammatical to a grammatical mode. However,
adults retain access to the pre-grammatical mode which they employ when
the conditions are appropriate. Givon also argues that others aspects of
language—such as the historical evolution of languages and creoliza-
tion—are also characterized by the same process of syntactization.

Pre-grammatical vs. grammatical discourse processing

Properties Grammatical mode Pre-grammatical mode

STRUCTURAL:
a. Grammatical morphology abundant absent
b. Syntactic constructions complex/embedded simple/conjoined
c. Use of word-order grammatical (subj/obj) pragmatic (topic/comment)
d. Pauses fluent halting
FUNCTIONAL:
e. Processing speed fast slow
f. Mental effort effortless laborious
g. Error rate lower higher
h. Context dependence lower higher
COGNITIVE:
i. Processing mode automated attended
j. Acquisition late early
k. Evolution late early

Table9.3: Pre-grammaticalv. grammatical discourse processing (from Givonl995)

A number of studies have tested Givon’s claims regarding syntactization
(for example, Schumann 1987; Givon 1984; Sato 1988; Pfaff 1992; Ramat
1992). They have produced mixed results. For example, Sato’s ten-month
study of two child Vietnamese learners of English found little evidence of
paratactic speech (non-propositional utterances, vertical constructions, and
scaffolded utterances). In fact, both learners, contrary to the theory, encoded
plenty of simple complete propositions from the start and were able to do so
without the help of interlocutor scaffolding. However, there was some evid¬
ence of a shift from coordination to subordination in one of the learners, al¬
though simple juxtapositioning of propositions served as the main way of
expressing logical relationships for both learners throughout the period of the
study. Clearer evidence of the absence of syntactization was seen in the learn¬
ers’ failure to produce infinitival complements until near the end, and the al¬
most complete absence of relative clauses and gerundive complements. Sato
suggests that interaction may be insufficient to ensure full syntactization, and
that encounters with written language may be crucial.
Somewhat more convincing evidence of syntactization comes from studies
that have investigated the hypothesis that grammatical markers develop out
of independent lexical items. Pfaff (1992) provides evidence for this in the ac¬
quisition of L2 German by pre-school and early school-age Turkish children
 Cognitive accounts of second language acquisition 373

in Berlin-Kreuzberg (the KITA Study). She found, for instance, that main verb
use of ‘sein’ and ‘haben’ preceded the auxiliary use of the same verbs. Ramat
(1992), in a study of the L2 acquisition of Italian by subjects of mixed lan¬
guage backgrounds, also provides support for this hypothesis. Adverbs pre¬
ceded the use of inflected modal verbs as a device for conveying various
meanings.
Givon has also identified a number of linguistic universals which are func¬
tional in nature. An example is the ‘quantity universal’, which states:

More continuous, predictable, non-disruptive topics will be marked by less

marking material; while less continuous, unpredictable/surprising, or dis¬
ruptive topics will be marked by more marking material (1984:126).

Pfaff’s (1987a) study of Turkish adolescents’ German interlanguage suggests

that this principle operates in L2 acquisition. Using data from two elicited
narratives, Pfaff was able to show that the protagonists in the stories were
almost always indicated by a pronominal form (‘less marking material’),
whereas all the other participants were marked by article + noun (‘more
marking material’). However, Tomlin (1990) reports a study of advanced L2
learners that did not show any alternation in the use of pronominal and nom¬
inal structures in accordance with the quantity principle; the learners gener¬
ally used nominal NPs throughout in a narrative task. Tomlin suggests that
the learners preferred nominal structures ‘as part of a communicative strategy
to ensure coherent and complete understanding by a listener’ (1990: 171).6
The status of functional universals like the quantity principle in L2 acquisi¬
tion is not yet clear.
Underlying both Klein’s and Givon’s positions is the assumption that learn¬
ers will be functionally motivated to develop their interlanguages. That is, the
drive to communicate more effectively leads learners to syntacticize. How¬
ever, Sato and others (for example, Higgs and Clifford 1982) have cast
doubts on whether communicative need by itself is sufficient to ensure high
levels of interlanguage development.

The Competition Model

Like other functionalist models,7 the Competition Model is a performance
not a competence model. That is, it seeks to account for the kind of know¬
ledge that underlies real-time processing in real-world language behaviour, al¬
though, as we will see, it has been investigated by means of experimental
studies which elicited rather artificial language responses. Also like other
functionalist models and unlike linguistic models associated with Universal
Grammar, the Competition Model sees the human capacity for language
learning as non-specific (i.e. as resulting from general cognitive mechanisms
involved in other kinds of learning). It also resembles other functionalist
models in that it is interactionist; that is, the learner’s grammar is viewed as an
374 Internal factors

emergent property resulting from the interaction between input and cognitive
mechanisms relating to perceptual abilities, channel capacity, and memory.
Central to the model is the idea of form-function mappings. As Mac-
Whinney, Bates, and Kligell (1984) put it:

The forms of natural languages are created, governed, constrained, ac¬

quired and used in the service of communicative functions.

Any one form may realize a number of functions and, conversely, any one
function can be realized through a number of forms. The learner’s task is to
discover the particular form-function mappings that characterize the target
language. In this respect, the model is close to some variability theories (for
example, Ellis 1985c; Huebner 1983).
Form-function mappings are characterized as being of varying ‘strengths’
in different languages. This is usually illustrated with reference to the func¬
tion of ‘agency’, which has a number of possible formal exponents:

1 Word order: in the case of transitive constructions, the first noun men¬
tioned in a clause is likely to function as the agent. For example, in the
English sentence ‘Mary kissed John’, ‘Mary’ is the agent.
2 Agreement: the noun phrase which functions as agent may agree in number
with the verb. Thus, in English, a singular noun phrase functioning as agent
takes a singular verb form (for example, ‘She likes ice-cream’), while a
plural noun phrase takes a plural verb form (for example, ‘They like
ice-cream’). The object of the sentence has no effect on the verb form.
3 Case: the noun phrase functioning as agent may be morphologically
marked in some way. For example, the agent is signalled in German by
nominative case marking on the article, while the object is signalled by
means of accusative case marking (for example, ‘Der Mann ifit den Apfel’ =
‘The man is eating the apple’).
4 Animacy: agents are normally animate, patients are normally inanimate.

Any one language is likely to utilize several devices for signalling the ‘agent’ of
a sentence. English, for instance, uses all four, as illustrated in these sentences:

Mary kissed John, (word order)

Money they like, (agreement)
She kissed him. (case)
This book Mary likes a lot. (animacy)

However, as these examples also show, a language is likely to assign different

weights to these devices in terms of the probability of their use in signalling a
given function. English, for example, relies primarily on word order to en¬
code agency, while Russian uses case marking and Japanese animacy. Like
variability models, the Competition Model is probabilistic in nature.
 Cognitive accounts of second language acquisition 37 5

The model take its name from the ‘competition’ that arises from the differ¬
ent devices or cues that signal a particular function. For example, in a sen¬
tence like ‘that lecturer we like a lot’ there is competition between ‘lecturer’
we’ and ‘lot’ for the role of agent of the verb. ‘Lot’ rapidly loses out because,
unlike ‘lecturer’ and ‘we’, it is inanimate, and because it follows rather than
precedes the verb. The candidacy of ‘lecturer’ is promoted by its position in
the sentence—it is the first noun—but, ultimately, this cue is not strong
enough to overcome two other cues. ‘We’ is the strongest candidate for agent
because it is nominative in case and because it agrees in number with the verb.
The task facing the L2 learner is to discover (1) which forms are used to
realize which functions in the L2, and (2) what weights to attach to the use of
individual forms in the performance of specific functions. This is what is
meant by ‘form-function mapping’. The input supplies the learner with cues
of four broad types: word order, vocabulary, morphology, and intonation.
The usefulness of a cue is determined by several factors: (1) ‘cue reliability’
(the extent to which a cue always maps the same form onto the same func¬
tion), (2) ‘cue availability’ (how often the cue is available in the input) and (3)
‘conflict validity’ (whether a cue ‘wins’ or ‘loses’ when it appears in competi¬
tive environments). For example, if we consider the information available to
the L2 learner regarding the role of word order in realizing agency in English,
we can characterize this ‘cue’ as relatively reliable (the noun phrase preceding
the verb is typically the agent) and readily available (the input is likely to sup¬
ply plentiful examples of this mapping). Also, in English, word order tends to
override other cues (except agreement). Thus, in a sentence like ‘Mary bit the
dog’, ‘Mary’ is the agent, even though experience of the world might lead one
to suspect that ‘the dog’ is the more likely agent.
There has also been some attempt to specify how learners use the informa¬
tion available from ‘cues’ to construct their language systems. McDonald
(1986) has proposed a learning-on-error model, according to which the
weights attached to specific form-function mappings are changed when the
learner interprets an input cue incorrectly and is subsequently provided with
feedback, a view of acquisition which, as we shall see later, accords with that
of Parallel Distributed Processing (see page 403). McDonald suggests that
this may account for the developmental shift noted in LI acquisition from an
initial dependence on cue reliability/availability to dependence on conflict
validity. To begin with, children respond to those cues which are salient and
easily detectable (such as word order), but once these have been established
they turn their attention to sentences containing conflicting cues (for
example, ‘That person we all love’, where there is conflict between the word
order and agreement cues). Ultimately it is these sentences that help them to
establish the dominance patterns of the cues. McDonald and Heilenman
(1991) provide evidence from an experimental study of French/English bilin¬
guals that supports such a view.
376 Internal factors

The Competition Model has informed a number of studies of L2 acquisi¬

tion (for example, Harrington 1987; Gass 1987; Kilborn and Cooremann
1987; Kilborn and Ito 1989; McDonald and Heilenman 1991; Sasaki 1991).
These studies take the form of sentence-interpretation experiments using bi¬
lingual subjects in a within-subjects, cross-language design. That is, speakers
of different languages are asked to identify the function of different cues in
both LI and L2 sentences that have been designed to reflect both the coordi¬
nation and competition of cues. For example, they may be asked to say which
noun is the agent of an action in acceptable sentences like ‘The boy is chop¬
ping the log’ and in semantically unlikely sentences such as ‘The logs are
chopping the boy’, where the animacy cue is in competition with the word
order cue, but the agreement cue is in coordination. The studies then compare
the responses of learners with different language backgrounds.
A good example of such a study is Harrington (1987). Harrington investig¬
ated the effects of three factors—word order, animacy and stress—on the
processing strategies used by native speakers of English (the NL English
group), native speakers of Japanese (the NL Japanese group), and Japanese
speakers of English (the interlanguage (IL) group). There were twelve adult
subjects in each group. They were asked to interpret 81 test sentences by stat¬
ing which of two nouns in each sentence was the logical subject (i.e. agent).
The sentences had been designed to incorporate the three factors systematic¬
ally, either in competition or in coordination. The subjects, however, were
not informed of this. Inevitably, some of the sentences were ungrammatical
or unnatural, but Harrington argued that this was appropriate and that, in
any case, information regarding the effects of converging and competing cues
was ‘extraordinarily difficult to obtain from natural speech’ (1987: 360).
The analysis of the results was based on ‘choice’ (which noun each subject
chose in each sentence) and ‘latency’ (the time it took each subject to make
each choice). The results are extremely complex, reflecting a number of inter¬
actions between the subject and language variables. We will focus here on the
results obtained for word order cue effects, which are summarized in Table
9.4. As expected, the NL English group interpreted the first noun in canonical
NVN sentences and the second noun in the unnatural NNV and VNN sen¬
tences8 as the agent. Also, as expected, the NL Japanese group demonstrated
little sensitivity to word order as a signal of agency, exhibiting no more than a
slight tendency to choose the first noun as the agent and no use at all of a ‘se¬
cond noun strategy’. In contrast to these results for the word order cue, the
NL English showed less sensitivity to the animacy cue than the NL Japanese,
although to complicate matters, a sub-group of the NL English did appear to
be influenced strongly by this cue. The IL group were midway between the NL
English and NL Japanese groups in their choice of agent in canonical NVN
sentences, but, as predicted, behaved like the NL Japanese group in their in¬
terpretation of the unnatural NNV and VNN sentences. Harrington also pro¬
duces results to indicate that even in the case of the NVN sentences, the IL
 Cognitive accounts of second language acquisition 377

group were responding more to the animacy than to the word order cue. He
concludes that a word order strategy was evident but was ‘only of limited
strength’ in the IL group. In contrast, there was clear evidence that animacy
cues were of primary importance in the IL group’s responses.

Group Canonical sentences Unnatural sentences

NVN NNV VNN

NL English 81% 35% 33%

NLJapanese 59% 56% 54%
IL subjects 68% 59% 56%

Table 9.4: Choice of first noun as agent in a sentence interpretation task (based on
results reported in Harrington (1987))

The results of this study are encouraging for the Competition Model. They
demonstrate that L2 learners are influenced by their LI processing strategies,
which they transfer when interpreting L2 sentences. Furthermore, the results
suggest that the processing strategies utilized by L2 learners can be located
somewhere on the continuum between the strategies used by native speakers
of the two languages concerned. This suggests that the idea of a restructuring
continuum may after all make some sense in the case of syntax, at least where
sentence processing is concerned. Further evidence for this is available in Kil-
born’s (1987) study of German learners of L2 English, which shows that a
group of advanced learners were much closer to a NL English group than
were a group of novice learners where word order cues were concerned. It
should be noted, however, that restructuring involves strategies rather than
forms; it is not L2 forms that replace LI forms, but rather L2 processing
strategies.
There is also evidence to suggest that some interpretation strategies are
more universal than others. Thus, Gass (1987) found that while English-
speaking learners of L2 Italian made little use of their LI syntax-based strat¬
egy, Italian-speaking learners of L2 English did transfer their lexical-semantic
strategy (i.e. animacy cues). Further evidence of similar directionality effects
in the transfer of LI interpretation strategies comes from Sasaki’s (1991) bi¬
directional study of the acquisition of English and Japanese. These results
testify to the primacy of a semantics/pragmatics strategy and confirm the find¬
ings of other functionally-orientated models of L2 acquisition. The utiliza¬
tion of LI interpretation strategies, therefore, depends on the universality of
the strategy; the more universal a strategy is, the more it is likely to be trans¬
ferred. Once again, then, we see evidence of constraints on transfer (see Chap¬
ter 8).
The strength of the Competition Model is that it allows researchers to test
very precise hypotheses. In this respect it resembles UG-based theories (see
Chapter 10). It also provides a convincing account of a number of aspects of
378 Internal factors

L2 acquisition which any theory must consider: the role of the LI, the effect of
input and the gradual way in which native-like ability is acquired. There are,
of course, other aspects which it does not address, at least at the moment. It is
not clear, for instance, what kind of knowledge learners use in sentence
interpretation, nor does the model have much to say about the cognitive
mechanisms responsible for obtaining intake from input or for using L2
knowledge in production.
Probably the main weakness of the model is over-reliance on rather artifi¬
cial interpretation tasks, a problem that is aggravated by the unnatural sen¬
tences that figure in such tasks. The justification for such a methodology is the
Ecological Validity Hypothesis, according to which ‘the processing of both
grammatical and ungrammatical sentences proceed by reference to the same
set of cues and processing patterns’ (MacWhinney, Pleh, and Bates 1985:
199). But this has been queried by McLaughlin and Harrington (1989):
... it may be wise to question the ‘ecological validity’ of an experimental
procedure in which subjects have to make decisions about sentences that
are as deviant in English as: ‘The apple is eating the man’. Perhaps subjects
are not processing such sentences as they would in actual communicative
situations, but are settling on a particular problem-solving strategy to get
them through the many judgements of this nature they have to make (1989:
125).
It might be further argued that L2 acquisition takes place as a result of ‘utter¬
ance processing’ rather than ‘sentence processing’, the distinguishing feature
being that utterances are contextualized whereas sentences are not. Utterance
processing involves pragmatic procedures, which are ignored in the kind of
sentence-processing tasks on which the Competition Model has relied.9
However, as the quotation from McLaughlin and Harrington suggests,
researchers are aware of the need to develop more natural, on-line ways of
investigating input processing.

The Nativization Model and Operating Principles

Whereas the Competition Model was designed to account for sentence inter¬
pretation, Operating Principles have been formulated to explain why certain
linguistic forms typically appear in learners’ (LI and L2) production before
others. The idea of operating principles is not incompatible with the Com¬
petition Model, however, as they shed light on the general ideas of ‘cue reliab¬
ility’ and ‘cue availability’. They are based on the general assumption that
those features that are easily attended to and easily processed will be the first
to be learnt and thus to be used in production.
Operating principles in L2 acquisition have been investigated by Andersen.
Andersen’s earlier work (Andersen 1979; 1980; 1983b; 1984a) sought an ex¬
planation for how learners create and restructure their interlanguage systems
as a product of participating in verbal interaction with more proficient speak-
 Cognitive accounts of second language acquisition 379

ers. It resulted in the ‘Nativization Model’, according to which L2 acquisition

consists of two general processes, nativization and denativization. In the for¬
mer, learners make the input conform to their own internalized view of what
constitutes the L2 system. That is, they simplify the learning task by forming
hypotheses based on knowledge that they already possess (LI knowledge and
knowledge of the world). In Andersen’s terms, they attend to an ‘internal
norm’. The result is the kind of pidginization evident in early language ac¬
quisition and documented in Schumann’s work (see Chapters 3 and 6). In
denativization, learners accommodate to an ‘external norm’; that is, they ad¬
just their interlanguage systems to make them fit with the input, making use
of inferencing strategies. Denativization is apparent in depidginization (the
elaboration of a pidgin language which occurs through the gradual incorp¬
oration of forms from an external source). Subsequently, Andersen (1990)
has recognized that nativization and denativization are not two separate
‘forces’ but aspects of the same overall process of acquisition. Andersen’s
later work is an attempt to develop the nativization model by further spe¬
cifying the ‘processes, cognitive operating principles, and communicative
strategies’ (1990: 48) that fit within it.
Andersen has been strongly influenced by Slobin’s idea of operating prin¬
ciples. In a series of publications culminating in the two-volume publication
The Crosslinguistic Study of Language Acquisition (1985a), Slobin has at¬
tempted to describe the universal principles that guide children in the process
of LI acquisition. These principles are conceived as the operating principles
by which children extract and segment linguistic information in order to
build a grammar of the language they are learning. According to Slobin
(1973; 1985b), children possess a ‘language making capacity’ consisting of
principles that enable them to perceive and segment items in the input (for ex¬
ample, ‘Pay attention to stressed syllables in extracted speech units. Store
such syllables separately and also in relation to the units in which they occur’)
and also principles that govern how they organize and store new information
(for example, ‘Keep track of the frequency of occurrence of every unit and
pattern that you store’). In his 1985b paper, Slobin identifies a total of 40 such
principles, a considerable increase on his earlier (1973) paper. Their invest¬
igation requires the painstaking analysis of production data from many unre¬
lated languages, as this is the only way to separate out language-specific
tendencies from universal principles.
Andersen points out that most of the operating principles he has identified
in his own work can be related to Slobin’s. However, he claims that his prin¬
ciples are not simply a translation to L2 learning, but rather ‘macro¬
principles’, each one of which corresponds to a group of principles in Slobin’s
framework. The seven operating principles described in Table 9.5 evolved
out of research that Andersen has conducted on the L2 acquisition of English
and Spanish (for example, Andersen 1990).
380 Internal factors
Principle
The one-to-one principle
The multi-functionality
principle
Formal determinism
principle
Distributional bias principle
Definition
‘An interlanguage system
should be constructed in
such a way that an intended
underlying meaning is
expressed with one clear
invariant surface form (or
construction)’ (Andersen
1984a: 79)
(a) Where there is clear
evidence in the input that
more than one form marks
the meaning conveyed by
only one form in the
interlanguage, try to discover
the distribution and
additional meaning (if any) of
the new form.
(b) Where there is evidence
in the input that an
interlanguage form conveys
only one of the meanings
that the same form has in the
input, try to discover the
additional meanings of the
form in the input.
Pay closer attention to form¬
meaning relationships that
are clearly and uniformly
encoded in the input than to
other form-function
relationships.
If both X and Y can occur in
the same environments A
and B, but a bias in the
distribution of X and Y
makes it appear that X only
occurs in environment A and
Y only occurs in environment
B, when you acquire X and
Y, restrict X to environment A
and Y to B.
Example
Clitic pronouns in French
and Spanish are placed
before the verb but in IL they
are placed after the verb, like
full NPs.
Spanish iearners of 12
English typically acquire a
single negator, ‘no’, to begin
with. However, the input
supplies evidence of other
negators, e.g. ‘not’ and
‘don’t’, each with a different
meaning.
Learners pay attention to
negators other than ‘no’ in
the input because the other
forms are modelled clearly in
the input such that their
meanings are transparent.
In Spanish, punctual verbs
tend to occur in the preterite
form, and state verbs in the
imperfect form; L2 learners
manifest this bias in the use
of the two tenses.
 Cognitive accounts of second language acquisition 381

Principle Definition Example

Relevance principle
Transfer to somewhere
principle
Relexification principle
If two or more functors apply
to a content word, try to
place them so that the more
relevant the meaning of the
functor is to the meaning of
the content word, the closer
it is placed to the content
word. If you find that a notion
is marked in several places,
at first mark it only in the
position closest to the
relevant content word.
A grammatical form or
structure will occur
consistently and frequently
in IL as a result of transfer if,
and only if: (1) natural
acquisitional principles are
consistent with the LI
structure, or (2) there already
exists within the L2 input the
potential for (mis-)
generalization from the input
to produce the same form or
structure.
When you cannot perceive
the structural pattern of the
L2, use your LI with lexical
items from the L2.
In the Spanish verb system,
aspect is most relevant to
the lexical item it is attached
to (i.e. the verb), tense has a
wider scope but is still
closely related to the verb,
and subject-verb agreement
is least attached to the verb.
L2 learners acquire aspect,
tense, and agreement in this
order.
French learners of L2 English
do not place pronouns
before the verb even though
this is possible in French
because no model for such
transfer is available in the
input.
Japanese learners of L2
English have been observed
to use English lexis in SOV
sentence frames, but this
may be short-lived because
English provides no
evidence of SOV word order.

Table 9.5: Operating principles in L2 acquisition (summarized from Andersen

1990)

Operating principles have been criticized in both LI and L2 acquisition re¬

search on a number of grounds. Dulay and Burt (1974d) and Larsen-Freeman
(1975) have argued that they are difficult to test and are not mutually exclus¬
ive. Also, it is not clear how many principles are needed to explain acquisi¬
tion. These criticisms, which were levelled at the earlier list (Slobin 1973), are
still valid. Bowerman (1985) points out that Slobin only lists operating prin¬
ciples, but this is inadequate because we need to know how the various
principles are related to each other and, importantly, what weight is to be
attached to principles when they conflict. Andersen acknowledges these cri¬
ticisms, but argues that they do not warrant rejecting the whole idea of
382 Internal factors

operating principles. The solution is to work on them and refine them.

However, as Pleh (1990) notes in her review of Slobin (1985a), there is an
overriding difficulty—so far, there is no explanation of the principles them¬
selves. Only when this is available will problems concerning the number and
the form of the principles be resolvable.

The Multidimensional Model and processing operations

The Multidimensional Model resembles Andersen’s work on operating prin¬
ciples in a number of respects. First, it is based on the painstaking analysis of
naturally occurring learner speech. Second, it also sees the regularities in
learner language as the product of cognitive processes that govern the lin¬
guistic operations learners are able to handle. However, the model constitutes
a considerable advance on the idea of operating principles in that it relates the
underlying cognitive processes to stages in the learner’s development, ex¬
plaining how one stage supersedes another. Also, the model provides an ac¬
count of inter-learner variation.
The Multidimensional Model was developed by a group of researchers in¬
volved in the Zweitspracherwerb Italienischer und Spanischer Arbeiter
(ZISA) project in the 1970s (see Clahsen, Meisel, and Pienemann 1983). This
involved the longitudinal study of 12 adults and the cross-sectional study of
45 adults, the researchers arguing that cross-sectional studies only had valid¬
ity if they were supported by longitudinal research. Subsequently, the model
has been developed further as a result of studies of the naturalistic L2 acquisi¬
tion of Australian English by Polish and Vietnamese immigrants (Johnston
1985) and also of L2 Japanese by (Yoshioka and Doi 1988). In addition,
some of its claims have been tested in classroom studies investigating the ef¬
fects of formal instruction on the acquisition of specific grammatical struc¬
tures (for example, Pienemann 1984; 1989).
The model is in several respects more comprehensive than any of the other
models so far considered, with the possible exception of Krashen’s Monitor
Theory. It makes the following general claims:
1 Learners manifest developmental sequences in the acquisition of a number
of grammatical structures, such as word order and some grammatical
morphemes.
2 Learners also display individual variation, both with regard to the extent to
which they apply developmental rules and to the extent to which they
acquire and use grammatical structures that are not developmentally
constrained.
3 Developmental sequences reflect the systematic way in which learners
overcome processing constraints. These constraints are of a general cog¬
nitive nature and govern production.
4 Individual learner variation reflects the overall orientation to the learning
task, which in turn is the product of socio-psychological factors.
 Cognitive accounts of second language acquisition 383

5 Formal instruction directed at developmental features will only be suc¬

cessful if learners have mastered the prerequisite processing operations
associated with the previous stage of acquisition. However, formal in¬
struction directed at grammatical features subject to individual variation
faces no such constraints.

The account of the Multidimensional Model that follows will consider points
(1) to (4). Point (5) is considered in Chapter 14.
The claim that learners acquire a number of grammatical features in a clear
developmental sequence was based, initially, on the detailed study of German
word order rules by the ZISA researchers. The general pattern is described in
Chapter 3 (see Table 3.8). Subsequent work has demonstrated that similar
developmental patterns are evident in the acquisition of L2 English word or¬
der rules. More importantly, the claim that learners follow developmental se¬
quences has also been extended to grammatical morphology, thus vastly
extending the scope of the model. Pienemann and Johnston (1988) distin¬
guish ‘local morphemes’ and ‘non-local morphemes’ according to whether
the functional mapping performed by the morpheme occurs within a major
constituent or across constituents. For example, the plural -s and the regular
past tense -ed morphemes are seen as ‘local’ because they involve simple addi¬
tions to the constituents to which they belong. However, 3rd person -s is
‘non-local’ because the function it performs (marking subject-verb agree¬
ment) involves the learner in relating two constituents, the subject noun
phrase and the verb. The extended sequence of acquisition of developmental
rules for English can be found in Table 3.9.
The ZISA researchers noted that within the defined stages of development,
there was considerable variation among learners (Meisel, Clahsen, and Piene¬
mann 1981). This variation is of two kinds. First, learners differ in the extent
to which they apply a particular word order rule to different linguistic con¬
texts. For example, the inversion rule in German applies in a number of con¬
texts—after an interrogative pronoun, after a preposing adverbial, after the
topicalization of a direct object, and after a sentence-initial adverbial
clause—but when learners acquire this rule they do not invariably apply it to
all these contexts. Meisel, Clahsen, and Pienemann (1981: 126) comment:
‘whereas some learners seem to acquire a rule perfectly before they move on
to the next learning task, others are less perfect in terms of the target lan¬
guage’. Second, learners vary in the extent to which they use restrictive simpli¬
fication and elaborative simplification. This distinction works in much the
same way in learner language as it does in foreigner talk (see the discussion in
Chapter 6). Thus, restrictive simplification is designed to achieve ‘optimal
results in communication’ by ‘reducing the grammar in a way that makes it
easy to handle’ (Meisel 1980: 36). It is always evident in the early stages of ac¬
quisition. Elaborative simplification is ‘a strategy that helps to complexify the
grammatical system’ by formulating hypotheses that are ‘approximations to
384 Internal factors

the actual rule’ and often involve overextensions of a rule (op. cit.: 37). Not
all learners engage in elaborative simplification.
The model therefore has two principal axes, the developmental and the
variational, as shown in Figure 9.4. This allows for learners to be grouped
both in terms of their stage of development and in terms of the kind of simpli¬
fication they engage in. For example, Figure 9.4 shows two learners (A and B)
at stage 5, two at stage 4 (C and D), and 2 others at stage 3 (E and F), reflecting
their progress on the developmental axis. It also shows differences between
the learners at each level. For example, learner B produces more standard-like
language than learner A. Meisel, Clahsen, and Pienemann (1981) emphasize
that progress on one axis or dimension is independent of progress on the
other. It is theoretically possible, therefore, for a learner who practises elab¬
orative simplification to use more target-like language overall than a learner
who is ‘developmentally’ far more advanced. Learner F, for instance, is only
at stage 3, but uses more standard-like constructions than Learner A, who is
at stage 5.

na

'x
(0

s 5 A B
c
CD
4 C D
E
Q.
O 3 E
CD
>
CD 2
"0

1
TL features
variational axis
Figure 9.4: The two axes of the Multidimensional Model

The identification of developmental patterns is of no value to theory-

construction unless a principled explanation for them can be found. It is in
this respect that the Multidimensional Model offers most to SLA research. It
explains why learners pass through the stages of development they do, and
also, on the basis of this, affords predictions regarding when other grammat¬
ical structures (those that have not yet been investigated) will be acquired. In
other words, the model has both explanatory and predictive power.
The explanation rests on the idea of processing constraints. Clahsen (1984)
has proposed that the sequence of acquisition of German word order results
from ‘reorderings and restructurings of various levels of underlying linguistic
units’ (1984: 221) and draws a direct comparison between his proposals and
Slobin s operating principles. He identifies three language processing strate¬
gies, each one corresponding to the degree of psychological complexity in¬
volved in the production of a particular word order rule:
 Cognitive accounts of second language acquisition 385

1 Canonical Order Strategy (COS)

No permutation or reordering of constituents in a structure occurs; utter¬

ances manifest a ‘basic’ order that reflects a direct mapping of meaning on to
syntactic form. This strategy blocks the interruption of the underlying
structure.

2 Initialization/Finalization Strategy (IFS)

No permutation involving the movement of an element from within a struc¬

ture occurs, but movement of an initial element in a structure to final position
and vice versa is possible.

3 Subordinate Clause Strategy (SCS)

No permutation of any element in a subordinate clause is possible, but move¬

ment of an element from within a main clause to another position in the
clause is possible. Progress entails the removal of the processing ‘blocks’. For
example, learners who have acquired the verb separation rule and so can pro¬
duce utterances like:

alle kinder muss die pause machen

(all the children must the pause make)

but have not yet acquired either inversion or verb-end (see Table 3.8 for ex¬
amples) have overcome the constraint of the Canonical Word Order Strategy
but not of the other two strategies. The three strategies are hierarchical in that
SCS is not manifest until IFS is evident, which in turn cannot be employed un¬
til COS has been accessed. The acquisition of grammatical structures occurs
as learners systematically overcome the constraints identified in these pro¬
cessing strategies. The strategies are themselves developmental, not in the
sense that learners do not already ‘know’ them—they clearly do, as they are
involved in LI production—but in the sense that they can only be accessed in
the L2 incrementally.
Johnston (1986) has described the general pattern of acquisition evident in
L2 learners of English in Australia and identifies the nature of the processing
operations involved at each of six stages of development, drawing on
Clahsen’s three strategies. The operations are summarized in Table 9.6 with ex
amples from L2 English. It should be noted that the mastery of each
processing operation is seen as making the acquisition of not one but several
structures possible.
The Multidimensional Model also has a socio-psyschological dimension
that accounts for the simplification strategies employed by individual learn¬
ers. This is described in terms of a continuum reflecting learners’ orientations
towards the learning task (Meisel, Clahsen, and Pienemann 1981). At one
end is segregative orientation, which arises when there is a lack of interest
386 Internal factors

Stage Processing operation Linguistic realization

1 Production relies on non-linguistic Undifferentiated lexical items; formulas

processing devices. Learner has no such as ‘I don’t know’ and ‘I can’t’.
knowledge of syntactic categories.
2 Production of simple strings of elements Canonical word order; intonation
based on meaning or information focus. questions (e.g. ‘You playing football?’).
Learner still has no knowledge of
syntactic categories.
3 Learner is able to identify the beginning Adverb-preposing (e.g. Today I play
and end of a string to perform football’); do-fronting (e.g. ‘Do you play
operations on an element in these football?’); neg + V (e.g. ‘No play
positions e.g. learner can shift an football’).
element from beginning to end of string
and vice versa. These operations are still
pre-syntactic.
4 Learner is able to identify an element Yes/No questions (e.g. ‘Can you
within a string and to move this element swim?’); pseudo-inversion (e.g. ‘Where
from the middle of the string to either the is my purse?’).
beginning or the end. This operation is
again characterized as pre-syntactic.
5 Learner is now able to identify elements wh-inversion (e.g. ‘Where are you
in a string as belonging to different playing football?’); Internal neg. (e.g. ‘He
syntactic categories. She is able to shift did not understand?’).
elements around inside the string.
6 Learner is now able to move elements Q-tags (e.g. ‘You’re playing football
out of one sub-string and attach it to today, aren’t you?’); V-complements
another element. This stage is (e.g. ‘He asked me to play football?’).
characterized by the ability of the learner
to process across as well as inside
strings.

Source: Based on Johnston (1986); Pienemann, Johnston, and Brindley (1988)

Table 9.6: Processing operations involved in the acquisition of grammatical rules

in contact with native speakers, discrimination on the part of native speakers,

and a general lack of either instrumental or integrative motivation. At the
other end is integrative orientation, which is evident in learners with either a
strong desire to assimilate into the L2 culture or in those with no desire to
acculturate but with a strong instrumental need to learn the L2. The socio-
psychological dimension of the model echoes the social theories of L2 acquis¬
ition considered in Chapter 6. It explains the extent to which individual learn¬
ers conform to target-language norms. The learner’s orientation, however,
has no effect on the general sequence of progression along the developmental
axis.
As we have already noted, the Multidimensional Model is powerful not
only because it provides a satisfactory explanation of observed development
 Cognitive accounts of second language acquisition 387

in learner-language, but because it also constitutes a predictive framework.

The cognitive dimension afforded by the identification of underlying pro¬
cessing strategies and operations allows researchers to form hypotheses re¬
garding which grammatical structures will be acquired at which general stage
of development. The predictive framework has been applied to Japanese to
test the validity of hypotheses relating to the acquisition of the particles ‘wa’,
‘ga’, and ‘o’ in a cross-sectional study. The results support the general predic¬
tion that the order of acquisition is wa > o > ga (Yoshioka and Doi 1988).10
Pienemann and Johnston (1988) also report on the successful testing of
predictions relating to a number of morphological and syntactical rules. The
predictive power of the model is probably greater than that of any other model
of L2 acquisition, with the possible exception of the Competition Model.
There are, however, several problems with the model. As Larsen-Freeman
and Long (1991) note, there is no account of how or why learners overcome
the processing constraints. Two other problems are the difficulty in identi¬
fying formulaic chunks of language and of establishing a priori which fea¬
tures are variational and, therefore, not subject to the constraints that govern
developmental features. If a learner, supposedly at stage 2, produces an utter¬
ance like ‘Where does he live?’, which belongs to stage 5, to what extent does
this falsify the theory? It is very easy to immunize the theory by claiming that
such an utterance is a ‘formula’ and thus not subject to processing con¬
straints. But unless there exists a clear method for identifying such formulas
before the analysis proceeds, the theory runs the danger of becoming unfalsi-
fiable. A similar argument can be made for variational features. Unless we are
able to stipulate beforehand which features are variational and which devel¬
opmental, it is easy to dismiss any feature that fails to conform to the pre¬
dictive framework of the model on the grounds that it is variational. These are
serious problems, for which Larsen-Freeman and Long advance no solutions.
Another problem concerns the operational definition of ‘acquisition’.
Whereas the original research on which the Multidimensional Model was
based quantified all the features examined by indicating their overall propor¬
tion of suppliance in obligatory contexts (as in Meisel, Clahsen, and Piene¬
mann 1981: 112), Pienemann and his co-workers have subsequently
redefined acquisition in terms of ‘onset’ (i.e. the first appearance of a gram¬
matical feature). Hulstijn (1987), while accepting that the study of ‘onset’ is
legitimate in a model that emphasizes the importance of processing opera¬
tions, nevertheless feels that much of the work lacks rigour because Piene¬
mann ‘does not set quantitative or qualitative criteria to be met by the
learner’s production, in order to be considered as evidence for the operation
of a predicted processing strategy’ (1987: 14). The problem of defining ‘on¬
set’, of course, derives in part from the difficulty of identifying formulas and
variational features.
Finally, we should note that the Multidimensional Model only provides an
explanation of acquisition in terms of learner production. It tells us nothing
388 Internal factors

about how learners come to comprehend grammatical structures, nor does it

inform about how comprehension and production interact. In particular, the
theory does not address how learners obtain intake from input and how this is
then used to reconstruct internal grammars. It is in this respect that the theory
is most limited.

Skill learning models of second language acquisition

The models and theories we have looked at so far have drawn extensively on
constructs borrowed from cognitive psychology, but they have been specially
designed to account for language learning (as opposed to some other kind of
learning) and have taken ‘language’—either the target language or learner
language—as their starting point. The theories which we will consider in this
section originate in cognitive psychology; they explain L2 acquisition in
terms of general skill learning.

Anderson’s Adaptive Control of Thought (ACT) Model

Anderson’s Adaptive Control of Thought (ACT) Model11 (Anderson 1976;
1980; 1983) rests on the distinction between declarative and procedural
knowledge.12 While admitting that the distinction between the two types of
knowledge is not water-tight, Anderson (1976) characterizes the essential dif¬
ferences in the form of three assumptions:

1 Declarative knowledge seems to be possessed in an all-or-none manner,

whereas procedural knowledge seems to be something that can be partially
possessed.
2 One acquires declarative knowledge suddenly, by being told, whereas one
acquires procedural knowledge gradually, by performing the skill.
3 One can communicate one’s declarative knowledge verbally, but not one’s
procedural knowledge. (1976: 117)

Learning a language, like any other type of skill learning (for example, driving
a car or playing tennis), involves the development of procedures that trans¬
form declarative knowledge into a form that makes for easy and efficient
performance.
This transition of declarative to procedural knowledge takes place in three
stages. In the declarative stage information is stored as facts for which there
are no ready-made activation procedures. For example, we may be aware that
‘drowned’ consists of ‘drown’ and ‘-ed’, and yet be unable to produce
‘drowned’ correctly in conversation. The second stage is the associative stage.
Because it is difficult to use declarative knowledge, the learner tries to sort the
information into more efficient production sets by means of ‘composition’
(collapsing several discrete productions into one), and ‘proceduralization’
(applying a general rule to a particular instance). For example, the learner
 Cognitive accounts of second language acquisition 389

may have learnt ‘drowned’ and ‘saved’ as two distinct items, but may come to
realize that they can be represented more economically in a production set: ‘If
the goal is to generate a past tense verb, then add -ed to the verb’. This may
then serve as a general procedure for generating past tense forms, including
incorrect ones (such as ‘goed’). Anderson (1983) notes that errors are particu¬
larly likely during the associative stage. In the autonomous stage, in which
procedures become increasingly automated, the mind continues both to gen¬
eralize productions and also to discriminate more narrowly the occasions
when specific productions can be used. For example, the learner may modify
the past tense production set (above) so that it applies to only a subset of
verbs. At this stage the ability to verbalize knowledge of the skill can disap¬
pear entirely.
Anderson (1980 and 1983) discusses classroom L2 learning in the light of
the ACT model. He sees the kind of knowledge taught to the classroom learn¬
er as different from adult LI knowledge:

We speak the learned language (i.e. the second language) by using general
rule-following procedures applied to the rules we have learned, rather than
speaking directly, as we do in our native language. Not surprisingly,
applying this knowledge is a much slower and more painful process than
applying the procedurally encoded knowledge of our own language.
(1980:224)

However, Anderson sees the differences between LI and foreign language

learning as merely a question of the stage reached. Whereas LI learners al¬
most invariably reach the autonomous stage, foreign language learners typic¬
ally only reach the associative stage. Thus, although foreign language learners
achieve a fair degree of proceduralization through practice, and can use L2
rules without awareness, they do not reach full autonomy.
The ACT model is enormously complex. A single communicative act, for
instance, is likely to involve an elaborate sequence of interrelated ‘production
systems’. It is not possible to do justice to this complexity here. The central
points to grasp are the theoretical claim that learning begins with declarative
knowledge which slowly becomes proceduralized, and that the mechanism
by which this takes place is practice.

McLaughlin’s information processing model

The most detailed attempt to apply an information processing model to L2
acquisition can be found in the work of McLaughlin (McLaughlin 1978b;
1980; 1987; 1990a; McLaughlin, Rossman, and McLeod 1983). Although
McLaughlin’s ideas have developed somewhat over the years, it is possible to
identify a number of central theoretical tenets related to (1) the idea of pro¬
cessing limitations and (2) the need for restructuring.
390 Internal factors

McLaughlin draws heavily on research in cognitive psychology into in¬

formation processing (for example, Shiffrin and Schneider 1977). Learners
are limited in how much information they are able to process by both the na¬
ture of the task and their own information-processing ability. They are not
capable of attending to all the information available in the input. Some of it
becomes the object of focused or selective attention, while other parts are at¬
tended to only peripherally. In order to maximize their information pro¬
cessing ability, learners routinize skills. Initially a skill may be available only
through controlled processing, which McLaughlin, Rossman, and McLeod
(1983) describe as follows:
... not a learned response, but a temporary activation of nodes in a se¬
quence. This activation is under attentional control of the subject and,
since attention is required, only one such sequence can normally be con¬
trolled at a time without interference (1983: 139).
In contrast, automatic processing ‘involves the activation of certain nodes in
memory every time the appropriate inputs are present’. Routinization, there¬
fore, helps learners to reduce the burden on their information-processing ca¬
pacity. It occurs when they have the opportunity to practise controlled
processes. Routinization results in quantitative changes in interlanguage by
making an increasing number of information chunks available for automatic
processing.
Information-processing capacity is also extended in another
way—through restructuring. This allows for qualitative changes in in¬
terlanguage. These changes relate to both the way knowledge is represented
in the minds of learners and also the strategies they employ. Representational
change involves a shift from exemplar-based to rule-based representation.
McLaughlin (1990a: 118) gives the ‘classic’ example of past tense learning in
English (the U-shaped learning pattern described on page 303). Another
example of the general transition from exemplar-based to rule-based repres¬
entation is the change from formulaic speech to rule analysis (see Chapter
3).13 Restructuring is also facilitated by the flexible use of learning strategies.
In a series of experiments, which are considered in Chapter 12, McLaughin
and his associates were able to show that ‘expert’ language learners display
greater flexibility in restructuring rules, and are therefore able to avoid
making certain types of error. McLaughlin (1990a) considers that practice is
also important for restructuring.
McLaughlin’s information processing model provides a useful general ac¬
count of the language learning process. It is particularly helpful in that it helps
to reconcile one of the problems we identified in the discussion of the explicit/
implicit distinction. If explicit learning is characterized as involving focal at¬
tention and implicit learning as peripheral attention—a correlation that
McLaughlin, Rossman, and McLeod (1983:142) actually make—and if this
distinction is seen as distinct from the controlled/automatic distinction, then
 Cognitive accounts of second language acquisition 391

the nature of the interface between the different types of knowledge becomes
clearer. It is possible to argue (Ellis 1993a) that whereas there is a clear inter¬
face on the controlled/automatic dimension (i.e. controlled knowledge can
convert into automatic knowledge over time through practice), the interface
between explicit and implicit knowledge is constrained by learnability factors
(for example, the kind of strategies identified by Clahsen in the Multidimen¬
sional Model). In other words, automatizing knowledge does not lead to the
conversion of explicit into implicit knowledge but only to the automatization
of explicit or implicit knowledge. The continua intersect to produce different
kinds of L2 knowledge, as shown in Figure 9.5.
Controlled Automatic

A B
Rules and items that have Rules and items that have
been learnt formally through been learnt formally through
Explicit focal attention and that require focal attention and that can be
controlled processing in accessed rapidly and easily by
performance. means of automatic processes.

C D
Rules and items that have Rules and items that have been
been learnt implicitly through learnt implicitly through peripheral
Implicit peripheral attention and that attention and that can be accessed
require controlled processing rapidly and easily by means of
in performance. automatic processes.

Figure 9.S: L2 knowledge as intersecting continua

McLaughlin’s discussion of restructuring suggests that a third dimension is

needed to complete the model. The idea of progression from an exemplar-
based to a rule-based representation cannot be easily equated with the con¬
version of explicit into implicit knowledge (or vice-versa). It is best treated as
a separate dimension, relating most clearly to the way in which implicit
knowledge changes qualitatively over time.
In some other respects the information-processing approach, as currently
formulated, is rather limited. McLaughlin frequently talks of ‘practice’, for
example, but never offers a clear definition. Does he mean ‘opportunity to use
a process under normal operating conditions’ (i.e. in communication), or ‘op¬
portunity to practise specific rules and items in contrived drills and exercises’ ?
If the latter is intended, the model runs up against the problem that several
studies of formal instruction have shown that such practice does not always
result in improved automatization or in restructuring (see Chapter 14). The
model, in fact, struggles to account for why learners seem to acquire some
rules and items before others. Also, the notion of restructuring provides only
a very general account of the transitions that occur in interlanguage systems.
It would not be possible on the basis of McLaughlin (1990a) to formulate pre¬
cise hypotheses regarding which linguistic features are restructured when.
Nor does the model give precise information about the factors that cause
learners to restructure.
392 Internal factors

Summary

In this section we have examined a number of different theories and models of

L2 acquisition that seek to explain how learners construct and organize their
L2 knowledge. These theories vary in many respects but they are all ‘cognit¬
ive’ in nature in the sense that they view L2 acquisition as a mental process
involving gradual mastery of items and structures through the application of
general strategies of perception and production.
The framework for investigating L2 acquisition shown in Figure 9.1 sug¬
gests that a complete theory will need to account for three aspects of the ac¬
quisition process:
1 The relationship between input and knowledge (how learners attend to
linguistic features in input and how they integrate this ‘intake’ into their L2
knowledge systems).
2 The representation of L2 knowledge (how L2 knowledge is organized in
long-term storage, how it is influenced by LI knowledge and how different
types of knowledge, such as implicit and explicit, are related to each other).
3 The relationship between L2 knowledge and output (how L2 knowledge is
used in production and how this relationship influences the way in which
L2 knowledge is stored).
The theories discussed in this section tend to focus on just one or two of these
three aspects. Table 9.7 provides a general summary of this section in relation
to these three aspects.

Aspects of a theory of L2 acquisition Theories

1 Relationship between input and L2 Competition Model Operating Principles

knowledge
2 Representation of L2 knowledge
a Manner of representation Functional theories
Skill-building theories
b Role of the LI Interlanguage Theory
Competition Model
c Relationship between implicit and The Monitor Theory
explicit knowledge Bialystok’s Model of Second Language
Learning
Schmidt’s account of ‘consciousness’ and
‘noticing’
3 Relationship between L2 knowledge and The Monitor Theory
output Variability theories
Operating Principles
Multidimensional Model

Table 9.7: Theories of L2 acquisition in relation to general aspects of the

acquisition process
 Cognitive accounts of second language acquisition 393

Cognitive accounts of second language communication

In this section, we turn to theoretical accounts of the procedures that learners
employ when using their L2 knowledge in communication. The focus shifts
from how learners construct their L2 systems to how they use them in com¬
munication. Of course, as we have already seen and as Table 9.7 indicates, the
distinction between ‘acquisition of L2 knowledge’ and ‘use of L2 knowledge’
is not clear-cut in cognitive theories, as a general assumption is that the way in
which knowledge is stored reflects the way in which it is used. This is most
evident in the skill-learning models of Anderson and McLaughlin, where the
ideas of‘proceduralization’ and ‘automatization’ relate very obviously to the
idea of how knowledge is controlled in language use. Nevertheless, the dis¬
tinction is valid, for, as Schmidt (1992: 359) has pointed out, a distinction can
be drawn between ‘procedural knowledge’ (Anderson’s concern) and ‘pro¬
cedural skill’ (‘the performance aspect of actually doing something in real
time’). The focus of this section is on ‘procedural skill’ rather than ‘proced¬
ural knowledge’.
To account for ‘procedural skill’ it is necessary to explain how learners
make use of their existing L2 knowledge and also how they overcome the
problems that result from insufficient L2 knowledge or inability to access L2
knowledge. To address these we will consider two aspects of L2 communica¬
tion: (1) L2 speech planning, and (2) communication strategies.

Second language speech planning and the development of

‘procedural skill’
L2 speech planning was considered briefly in Chapter 4 in relation to variabil¬
ity in learner language. Here we will focus more on how learners’ ability to
plan their speech changes over time (the developmental aspect), drawing on
the work of a group of researchers at Kassel University, Germany (see Dech¬
ert, Mohle, and Raupach 1984). Their work is informed by Anderson’s ACT
model. Dechert (1984b: 216), for instance, notes that little attention has been
paid in SLA research to ‘procedural knowledge in language production’ and
argues that this is in need of special attention. It is clear, however, that he is
more concerned with ‘procedural skill’ than ‘procedural knowledge’. For
Dechert and the other Kassel researchers, the study of speech planning
phenomena serves as a means for examining the nature and extent of learners’
procedural skill. Mohle and Raupach (1989) suggest that one way of distin¬
guishing the application of declarative knowledge found in the associative
stage of development (characteristic of many advanced L2 learners) from the
full proceduralization evident in the autonomous stage (characteristic of
native-speaker speech) is by examining speech planning phenomena.
These speech planning phenomena are of two basic kinds; temporal vari¬
ables and hesitation phenomena. Table 9.8 summarizes the main variables, as
394 Internal factors

described in Wiese (1984). Both temporal variables and hesitation phenom¬

ena are on-line measures of speech and related to the idea of ‘fluency’. They
have been derived from the earlier work of psycholinguists such as Goldman-
Eisler (1968) and Butterworth (1980b).

Speech planning phenomena Description

A Temporal variables Variables related to the rate of speaking.

1 Speech rate The number of syllables per second. This
measure includes pause time.
2 Articulation rate The number of syllables per second of
time of articulation. This measure
excludes pause time.
3 Pause length The mean length of pauses above a stated
threshold level (e.g. 0.20 seconds).
4 Length of run The mean number of syllables between
pauses.
B Hesitation phenomena Variables relating to linguistic features that
disturb the smooth flow of speech.
1 Filled pauses The use of phonetic devices such as ‘uh’
or ‘mhm’ to fill pauses.
2 Repetitions The unchanged re-occurrence of a sub¬
string of an utterance that has no
syntactic/semantic function.
3 Corrections A change made to some part of the
preceding utterance—from a single
phoneme to a long sequence of text.

Table 9.8: Speech planning phenomena (derived from Wiese 1984)

The basic methodology employed by the Kassel researchers is to record L2

learners performing an oral task such as telling a story in their L2 and their LI
and, in many cases to obtain a native-speaker performance of the same task.
The recorded speech is then carefully transcribed indicating such features as
pause length, intonation contours, vowel lengthening, fillers, drawls, and
false starts. These are then analysed using both qualitative methods (the de¬
tailed discussion of sections of a text) and quantitative ones (measuring
speech rate etc.). The overall aim of work on speech planning phenomena is
to describe the processes learners employ in the production of L2 speech. As
an example of this work we will consider Raupach’s (1983) analysis of
formulas.
Through an analysis of the spontaneous productions of German students
of French from the Kassel corpus, Raupach shows that formulas function
both as ‘fillers’ and ‘organizers’. For example, in the case of the former, the
learners resort to such expressions as ‘je ne sais pas’ and single words such as
 Cognitive accounts of second language acquisition 395

mais and vraiment’ to fill pauses and, in so doing, give themselves time for
further planning activities. Organizers (for example, ‘je crois que ..and ‘on
peut...’) contribute to the development of ongoing speech by helping learn¬
ers to establish a structure for phrases and sentences. These formulaic chunks
provide learners with ‘islands of reliability’ that they can fall back on when
they experience planning problems.
The study of speech planning phenomena can shed light on the nature of
the development that learners undergo in acquiring procedural skill. As well
as acquiring L2 knowledge, learners need to increase their control over that
knowledge which has already been acquired (i.e. learn how to process this
knowledge in unplanned as well as planned language use). Towell (1987) re¬
ports on a four-year long study of a British student’s acquisition of L2 French,
based on interview-type recordings made three times a year. In order to in¬
vestigate the extent to which this learner’s ‘channel control’ developed over
time, Towell obtained measures of a number of temporal variables in approx¬
imately four minutes of speech elicited in the first and third years of the study
when the learner was telling the same anecdote. Towell reports that the learn¬
er increased her speaking rate by 65 per cent, her pause/time ratio by 37 per
cent, her articulation rate by 20 per cent, and the length of runs between
pauses by 95 per cent. Also the average length of utterances increased consid¬
erably. Lennon (1989) also reports the advances in overall ‘fluency’ made by
four advanced German learners of L2 English during a six-month residency in
England. In this case ‘fluency’ was measured by means of native speakers’ rat¬
ings of temporal variables.
One interesting possibility, discussed by Towell, is that the acquisition of
‘knowledge’ and ‘control’ (procedural skill) can proceed independently, with
learners often opting for one or the other. This idea was supported by my
study in Ellis (1990b). I examined the relationship between the development
of ‘control’ (measured in terms of speech rate) and the development of ‘know¬
ledge’ (measured in terms of the acquisition of word order rules and also by
means of standard proficiency tests) in 39 adult classroom learners of L2 Ger¬
man. I found that the correlations between the measures of control and
knowledge were very low, but that word order acquisition was significantly
and negatively correlated with gains in speech rate. One interpretation of this
result is that learners who opted to increase their store of L2 knowledge paid a
price in terms of procedural skill and vice versa. Schmidt (1992) also con¬
siders the acquisition of knowledge and procedural skill as potentially pro¬
ceeding separately. He gives as an extreme example a speaker of a pidginized
interlanguage who can speak it very fluently.
The study of speech planning phenomena in relation to the development of
L2 procedural skill is of obvious importance, but is also in its infancy. The
work to date helps to shed light on the status of the learners’ linguistic know¬
ledge, for example by providing evidence of the use of formulaic chunks, and
also, in longitudinal research, it helps to show in what ways learners develop
396 Internal factors

their procedural skill. One question, identified by Schmidt (1992) as the key
one, is the extent to which the development of procedural skill is dependent
on learners learning how to produce specific items or exemplars—as in the
case of formulas—or on learners learning how to perform abstract rules. It is
premature to reach any conclusion on this but as Schmidt notes, ‘there is ...
little theoretical support from psychology for the common belief that the de¬
velopment of fluency in a second language is almost exclusively a matter of
the increasingly skilful application of rules’ (1992: 377).

Communication strategies
We now turn our attention to the second aspect of procedural skill: the com¬
munication strategies (CSs) that learners use to overcome the inadequacies of
their interlanguage resources. CSs are used primarily to deal with lexical
problems, such as when a learner who does not know the word for ‘art gal¬
lery’ refers to it as a ‘picture place’. CSs can also be used to get around a gram¬
matical problem, as when a learner deliberately elects to use ‘ask’ instead of
‘make’ because of uncertainty regarding which form of the infinitive (plain in¬
finitive or ‘to’ + infinitive) to use with ‘make’.
As a starting point, it is useful to distinguish two broad theoretical ap¬
proaches to CSs. They can be viewed as discourse strategies that are evident in
interactions involving learners, or they can be treated as cognitive processes
involved in the use of the L2 in reception and production. If the first perspect¬
ive is taken, the discussion of CSs belongs properly to the study of learner in¬
teraction (see Chapter 7). This is, in fact, how Larsen-Freeman and Long
(1991) handle them. They treat them as devices of ‘conversation mainten¬
ance’. In general, however, SLAR has treated CSs as cognitive processes and it
is as a reflection of this that they are dealt with in this chapter.

The interactional approach

The study of CSs in an interactional approach began with Varadi (1980), in a
paper that circulated for many years before its publication. Tarone (1977)
also adopted an interactional approach. She defined CSs as involving ‘a mu¬
tual attempt of two interlocutors to agree on a meaning in situations where
requisite meaning structures do not seem to be shared’ (Tarone 1981: 419).
As such they differed from production strategies, which were defined as
attempts to ‘use one’s linguistic system efficiently and clearly, with a
minimum of effort’.
In her 1977 study, Tarone provided a typology of CSs, which is reproduced
in Table 9.9. Nine subjects were asked to describe two simple drawings and a
complex illustration in their L2 (English) and native language. The CSs
 Cognitive accounts of second language acquisition 397

described in Table 9.9 were derived from analysing transcripts of the learners’
attempts to refer to a number of objects and events depicted. The strategies,
therefore, were directly observable in the transcripts of the learners’ produc¬
tions. They are interactional—they reflect learners’ attempts to make them¬
selves understood to their interlocutors. Tarone’s methodology has served a
basis for subsequent studies of CSs, resulting in further typologies (see
Tarone, Cohen, and Dumas (1976), Varadi (1980), and Paribakht (1985)).

Communication strategy Description of strategy

1 Avoidance
a Topic avoidance
b Message abandonment
2 Paraphrase
a Approximation
b Word coinage
c Circumlocution
3 Conscious transfer
a Literal translation
b Language switch
4 Appeal for assistance
5 Mime
Avoiding reference to a salient object for which learner does
not have necessary vocabulary
The learner begins to refer to an object but gives up
because it is too difficult
The learner uses an item known to be incorrect but which
shares some semantic features in common with the correct
item (e.g. ‘worm’ for ‘silkworm’)
The learner makes up a new word (e.g. ‘person worm’ to
describe a picture of an animated caterpillar)
The learner describes the characteristics of the object
instead of using the appropriate TL item(s)
The learner translates word for word from the native
language (e.g. ‘He invites him to drink’ in place of ‘They
toast one another’)
The learner inserts words from another language (e.g.
‘balon’ for ‘balloon’). NB Subsequently, Tarone (1981) refers
to this as ‘borrowing’)
The learner consults some authority—a native speaker, a
dictionary
The learner uses a nonverbal device to refer to an object or
event (e.g. clapping hands to indicate ‘applause’)

Table 9.9: Tarone’s (1977) typology of communication strategies

Research based on typologies, such as Tarone’s, has focused on describing

the strategies used by different learners and on identifying the factors that in¬
fluence strategy choice (for example, the learners’ level of L2 proficiency, the
learners’ personality, the learning situation, and the nature of the task used to
elicit data). This research suffers from a number of problems, which have
been thoroughly documented by Bialystok and Kellerman (1987), Bialystok
(1990a), and Kellerman (1991). Bialystok notes that ‘the criteria for as¬
signing an utterance to a specific strategic category are sometimes vague,
sometimes arbitrary and sometimes irrelevant’ (1990a: 75) with the result
398 Internal factors

that the taxonomies are frequently not reliable. The taxonomies also lack va¬
lidity in that they are not generalizable across tasks and, most important, they
are not psychologically plausible. Kellerman (1991: 146) points out that re¬
ferring to ‘an art gallery’ as ‘a picture-place’ or as ‘a place where you look at
pictures’ clearly reflects the same underlying cognitive process. Therefore, to
code them separately as ‘word coinage’ and ‘circumlocution’, as happens in
Tarone’s taxonomy, is misleading. He argues that it is not necessary to posit
different strategies simply because they have different linguistic realizations.
Finally, the taxonomies are not parsimonious. Categories tend to proliferate
like the branches of an enormous tree.

The psycholinguistic approach

The psycholinguistic approach is illustrated by the work of Fserch and Kasper
(for example, Fasrch and Kasper 1980,1983a, 1983b, and 1985). CSs are loc¬
ated within a general model of speech production, in which two phases are
identified: a planning phase and an execution phase. The aim of the planning
phase is to develop a plan which can then be executed to allow the speaker/
hearer to achieve communicative goals. In this phase ‘the language user se¬
lects the rules and items which he considers most appropriate for establishing
a plan, the execution of which will lead to verbal behaviour which is expected
to satisfy the original goal’ (1983a: 25). CSs are seen as part of the planning
process. They are called upon when learners experience some problem with
their initial plan which prevents them from executing it. One solution is
avoidance. This occurs when learners change their original communicative
goal by means of some kind of reduction strategy. The other solution is to
maintain the original goal by developing an alternative plan through the use
of an achievement strategy. The typology of CSs which Fserch and Kasper
propose is based on this general distinction. Reduction strategies are further
subdivided into ‘formal’ (avoidance of L2 rules of which the learner is not cer¬
tain) and ‘functional’ (avoidance of speech acts, topics, and modality mark¬
ers). Achievement strategies are also divided into those that are
‘compensatory’ (replacement of an initial plan with a ‘strategic’ plan) and ‘re¬
trieval’ (perseverance with the initial plan by, for example, searching for the
item(s) required).
The perspective afforded by this model is clearly psycholinguistic, although
many of the categories in their typology resemble those found in Tarone’s.
For example, where Tarone has ‘appeal for assistance’, Fserch and Kasper re¬
fer to ‘co-operative strategies’ involving ‘direct’ and ‘indirect’ appeal. Fserch
and Kasper’s typology is an advance on that of Tarone, however, in that their
psycholinguistic framework provides a basis for classifying the CSs into cat¬
egories rather than just listing them. It locates the strategies they identify
within a theory of L2 production.
 Cognitive accounts of second language acquisition 399

In the Fasrch and Kasper model CSs are seen as ‘strategic plans’, which con¬
trast with ‘production plans’. Strategic plans differ from production plans in
two main respects: (1) problem-orientation and (2) consciousness. Learners
employ CSs because they lack the L2 resources required to express an inten¬
ded meaning (a problem in the planning phase) or they cannot gain access to
them (a problem in the execution phase). In either case, there is a lack of bal¬
ance between means and ends (Corder 1978c). CSs are, of course, not alone in
being problem-orientated, as learning strategies can also be motivated by
learners’ recognition of gaps in their L2 knowledge. CSs differ from learning
strategies in that the strategies are employed to meet a pressing communicat¬
ive need—they are short-term rather than long-term solutions to a problem
(Ellis 1985a: 181).
Fasrch and Kasper (1980) see consciousness as a secondary defining cri¬
terion of CSs. They recognize various problems in claiming that CSs are con¬
scious and, in an attempt to deal with these, distinguish plans that are (1)
always consciously employed, (2) never consciously employed, and (3) con¬
sciously employed by some but not all learners in some but not all situations.
They suggest that CSs are plans belonging to (1) and (3), i.e. they ‘are poten¬
tially conscious plans for solving what to an individual presents itself as a
problem in reaching a particular communicative goal’.
Both the problematicity and the consciousness criteria have been criticized.
Bialystok (1990a) argues that it is not clear how the distinction between ‘pro¬
duction plans’ (which are non-problematic) and ‘strategic plans’ (which are
problematic) manifests itself in actual language processing. She notes that
certain non-problematic instances of language use, such as giving definitions,
will result in exactly the same kind of overt linguistic behaviour associated
with CSs. For example, ‘an instrument for grating cheese’ might be con¬
sidered as a circumlocution for ‘cheese-grater’ in one context, but as a
straightforward definition of the object in another. Bialystok also criticizes
Fserch and Kasper’s claim that CSs are potentially conscious on the grounds
that there is no ‘independent means’ for deciding which plans fall into this
category and that without this ‘one is left to assume that all plans are poten¬
tially conscious’ (1990a: 5).
At the purely descriptive level the identification of CSs is not necessarily
problematic for as Bialystok notes, ‘strategies ... should reliably produce
forms of language that are different from those that one would expect to em¬
anate from the autonomous processing system’ (1990a: 19). Thus, they can
be identified by comparing performance of the same task in the LI and the L2,
although few studies have done this. It is at the psycholinguistic level that def¬
inition of CSs becomes difficult. According to Bialystok, it can only be
achieved if CSs are located within ‘a coherent account of speech production’
(1990a: 82). This, of course, is exactly what Faerch and Kasper have tried to
do but Bialystok, while approving of their attempt, considers it ultimately
inadequate because the basic concepts of ‘process,’ ‘plan’, and ‘strategy’ are
400 Internal factors

ill-defined and there is insufficient evidence to support the claim that planning
and execution can be distinguished in the way Faerch and Kasper do.
Bialystok’s own solution to the problem of providing a psycholinguistic
definition is to locate CSs within the theoretical framework that informs her
whole work, i.e. the distinction between ‘analysis’ (or knowledge) and ‘con¬
trol’. She distinguishes ‘knowledge-based’ and ‘control-based’ CSs. The for¬
mer involve the speaker in making some kind of adjustment to the content of
the message by exploiting knowledge of the concept, for example by provid¬
ing distinctive information about it, as in a definition or circumlocution. In a
control-based strategy, the speaker holds the initial intention constant and
manipulates the means of expression by integrating resources from outside
the L2 in order to communicate it, as in the use of many LI-based strategies
and mime. Bialystok emphasizes that both types of strategies need not be
characterized by problematicity (gaps in knowledge or communication
breakdown). Learners, like native speakers, select from their available op¬
tions in order to communicate their intentions as precisely as they can within
the constraints set up by particular tasks.
Bialystok’s solution to the problem of finding a psycholinguistic definition
is promising, as it enables CSs to be considered in relation to a general theory
of language processing based on a distinction (knowledge v. control) that is
firmly-grounded in cognitive psychology. Arguably, too, it provides a basis
for overcoming the objections that have been levelled against much of the re¬
search into CSs.

The Nijmegen project

The most extensive study of CSs to date is the Nijmegen project, reported in a
series of papers by Bongaerts, Kellerman, and Poulisse. The main study,
which we will consider here was that of Poulisse (1990a; 1990b). Other stu¬
dies (for example, Kellerman, Bongaerts, and Poulisse 1987; Kellerman,
Amerlaan, Bongaerts, and Poulisse 1990) have made use of the same theoret¬
ical framework but with different materials or methods.
Poulisse’s 1989 study, like the other studies, investigated lexical strategies
only, seen as referential in nature (i.e. they involve attempts by speakers to en¬
code messages that enable hearers to identify specific referents). The subjects
consisted of three groups of 15 Dutch learners of L2 English with varied profi¬
ciency, as indicated by the number of years they had been studying English,
school grades, teacher judgements, and cloze test scores. All the subjects were
asked to perform four tasks in English: (1) a concrete picture description task
involving everyday objects, the names of which the subjects were unlikely to
know in English, (2) an abstract figure description task, (3) an oral interview,
and (4) a story retelling task, where the subjects listened to a story in Dutch
and retold it in English with the help of picture prompts. (2) was also per¬
formed in the learners’ LI. (3) and (4) were video-recorded and played back
 Cognitive accounts of second language acquisition 401

to the subjects for their retrospective comments, which were audio-recorded

and transcribed.
CSs were defined as ‘strategies which a language user employs in order to
achieve his intended meaning on becoming aware of problems arising during
the planning phase of an utterance due to (his own) linguistic shortcomings’
(Poulisse 1990a: 88). Few difficulties were encountered in the identification
of CSs in the first two tasks, as the problems the learners would encounter had
been predetermined through the design of the tasks. Tasks (3) and (4) pre¬
sented greater difficulty, however, and to ensure reliability two independent
judges were asked to identify CSs, and the retrospective comments of the
learners were also used. Only ‘clear cases’ were finally included in the
analysis.
One of the main contributions of this study is a simple and elegant tax¬
onomy of CSs that the investigators claim has psychological plausibility (i.e.
it reflects the nature of the mental processing involved in the production of
CSs). This taxonomy is shown in Table 9.10. The taxonomy rests on two
archistrategies, labelled ‘conceptual’ and ‘linguistic’ (or ‘code’), which Keller-
man (1991) describes as follows:

Learners can either manipulate the concept so that it becomes expressible

through their available linguistic (or mimetic) resources, or they can ma¬
nipulate the language so as to come as close as possible to expressing their
original intention.
These archistrategies can be further broken down, as shown in Table 9.10,
and are eventually traceable to many of the CSs found in taxonomies such as
Tarone’s (1977). The distinctions between ‘analytic’ and holistic’ and be¬
tween ‘transfer’ and ‘morphological creativity’ constitute poles in a con¬
tinuum of options rather than discrete options.
The Nijmegen typology highlights the commonalities that underlie what
are discrete strategy types in Tarone’s framework. It reveals, for instance, that
word coinage, circumlocution, and approximation are ‘conceptual’ in
nature, and that they all involve an ‘analytic’ process. It also distinguishes
between types of non-verbal behaviour instead of lumping them together in a
single category, ‘mime’. Thus, ‘ostensive definition’ (pointing at an object) is
seen as the non-verbal equivalent of a linguistic strategy, whereas ‘mimetic
gesture’ (the modelling of some feature(s) of a referent through gesture, per¬
haps accompanied by imitative sounds) is ‘conceptual’ in nature.
The study examined the effects of proficiency and task on the learners’ use
of CSs and also considered whether there was any evidence that learners used
CSs differently from native speakers. The main findings were as follows. The
less proficient learners used more CSs than the more proficient, a function of
their more limited control of L2 vocabulary. There was also some limited
evidence of proficiency-related effects on the type of strategies used. For ex¬
ample, the more advanced learners made greater use of holistic strategies
involving superordinates. However, there were very few proficiency-related
402 Internal factors

Archistrategies Communication strategies

Conceptual 1 Analytic (circumlocution, description, and paraphrase)

2 Holistic (the use of a superordinate, coordinate, or subordinate
term)
Linguistic 1 Transfer (borrowing, foreignizing, and literal translation)
2 Morphological creativity

Table 9.10: The typology of communication strategies used in the Nijmegen

Project (based on Poulisse 1990; Chapter 7)

differences in the ways in which the different strategies were realized. On the
other hand, the nature of the task was found to have a marked effect on strat¬
egy selection. The subjects preferred elaborate analytic strategies in task (1),
while in tasks (3) and (4) they made greater use of short holistic strategies and
transfer strategies. Finally, the comparison of LI and L2 referential behavi¬
our failed to reveal any significant differences. The investigators have used
this finding to challenge the ‘uniqueness fallacy’, (the claim that CSs are a dis¬
tinctive second-language phenomenon), and to argue that L2 learners do not
have to develop a special L2 strategic competence but instead can apply their
LI strategic competence.

An evaluation of communication strategy research

Clearly, the Nijmegen taxonomy is a great improvement on the earlier taxo¬
nomies in that it locates the description of CSs within a parsimonious cognit¬
ive framework. In general, though, work on CSs has been limited in scope.
The research to date has concentrated almost exclusively on lexical problems
and sheds no light on the CSs that learners use to cope with gaps in their gram¬
matical knowledge. It is not clear whether the kinds of taxonomies that have
been developed can cope with grammatical problems.
Furthermore, little has been discovered about the developmental nature of
CSs in L2 production. Do learners simply use the strategic competence they
have developed in relation to LI production, or do they have to re-learn it for
L2 use? Bialystok’s (1983a) study suggests that there is a general switch from
Ll-based to L2-based strategies, but Poulisse (1990a) found little evidence of
proficiency-related effects on strategy choice. This is an important issue for
language pedagogy, as it bears on whether L2 learners need to be taught how
to use CSs or can be expected to transfer them from their LI. Kellerman
argues that ‘there is no justification for providing training in compensatory
strategies ... Teach the learners more language and let the strategies look
after themselves’ (1991: 158). But other researchers have felt that strategy
training is desirable (for example, Fserch and Kasper 1983a). At the moment,
 Cognitive accounts of second language acquisition 403

all that can be said is that there is insufficient evidence on which to base a
decision.
The research to date has also made no attempt to investigate the relation¬
ship between the use of CSs and acquisition. There is only speculation.
Corder (1978c) characterizes reduction strategies as ‘risk-avoiding’ and
achievement strategies as ‘risk-taking’, and suggests that the latter rather than
the former will contribute to successful language learning. Fserch and Kasper
(1980) express the same view, arguing that achievement behaviour encour¬
ages hypothesis formation, and also that risk is essential for automatization.
Tarone (1980b), however, holds a different view, suggesting that CSs of any
kind help learners to negotiate their way to the right target language forms.
Also, CSs in general keep the channel open and thus secure more input for the
learner. They are the means by which learners can act on Hatch’s (1978b) ad¬
vice never to give up. However, it could also be argued that CSs inhibit ac¬
quisition. Learners like Schmidt’s (1983) Wes have been found to develop
their strategic competence at the apparent expense of their linguistic compet¬
ence. These conflicting views can only be resolved through careful research.

Summary

In this section we have considered two aspects of L2 production. We have

found that where ‘procedural skill’ is concerned, there are clear differences
between learners and native speakers, as evident in temporal variables and
hesitation phenomena. Acquisition is characterized by a progressive control
of L2 resources leading to greater fluency. Furthermore, there is some evid¬
ence to suggest that this development can occur independently of the develop¬
ment of L2 linguistic knowledge. In the case of communication strategies, the
second aspect of L2 communication we have considered, the research has
concentrated on identifying the strategies that learners use and developing
taxonomies to account for them. The best taxonomy is that provided by the
Nijmegen project. This characterizes the cognitive nature of CSs in terms of
the distinction between ‘conceptual’ strategies, where learners engage in
some manipulation of the concept that they are trying to convey, and ‘lin¬
guistic’ strategies, where learners resort to their LI (or another L2) and to
morphological creativity. Little is currently known, however, about whether
CSs are developmental in nature, or of how their use affects L2 acquisition.

Parallel Distributed Processing and second language acquisition

The last account of L2 acquisition to be considered in this chapter is of a rad¬
ically different nature. Parallel Distributed Processing (PDP) differs from
other models in several respects but in particular in the way in which know¬
ledge and learning are conceptualized. Knowledge is seen not as ‘patterns’
or ‘rules’, nor is there any distinction drawn between ‘declarative’ and
404 Internal factors

‘procedural’ knowledge. Instead, what learners ‘know’ is characterized as a

labyrinth of interconnections between ‘units’ that do not correspond to any
holistic concepts of the kind we normally recognize (for example, a real-life
person or object or, where language is concerned, a word or grammatical fea¬
ture). Performance (i.e. information processing) involves the activation of the
requisite interconnections; learning arises when, as a result of experience, the
strength of the connections (i.e. ‘weights’) between units is modified. PDP
constitutes ‘a unified model of cognition, straddling the traditional compet¬
ence/performance distinction’ (Plunkett 1988: 307).
Although the idea of ‘parallel processing’ is not a new one in psychology,
the current interest in it has been stimulated by the recent work of Rumelhart,
McClelland, and the PDP Research Group (see Rumelhart, McClelland, and
the PDP Research Group 1986a and 1986b). This work entails the develop¬
ment of both a theory to explain information processing and learning, and
various computer models that allow simulations of actual information pro¬
cessing or learning tasks to be run. The computer models function as tests of
the theory. Both the theory and the models are extremely ambitious; as
Schmidt (1988) notes, they aim ‘to provide a mathematical model that cap¬
tures the essence of both neural processing, on the one hand, and thought on
the other’ (1988:55). They are also of great complexity, so no attempt will be
made to explicate or illustrate them fully here. The main constructs that in¬
form the theory will be considered, a brief account of one simulation of lan¬
guage learning will be provided, and some of the advantages and
disadvantages of PDP as an account of L2 acquisition will be examined.
A good starting point for trying to understand PDP is an example of the
kind of models of cognition that have been proposed. The one that follows is
taken from McClelland, Rumelhart, and Hinton (1986). The problem ad¬
dressed is ‘How do we remember someone we know?’ One possibility is that
we construct separate memory boxes for each person we know, creating a
kind of content index which we can consult when we want to recall a particu¬
lar individual. One of the problems of such a theory is that we cannot easily
locate the right entry in the index if some of the information we are using is
faulty or incomplete. A second possibility—the one preferred by McClelland,
Rumelhart, and Hinton—is that the information relating to a particular per¬
son is spread over a number of inter-connected units, such that the activation
of one unit leads to the activation of all the other units, so making the total
contents of the memory for this person available. In this case, the units consist
of properties such as ‘name’, ‘age’, ‘education’, ‘marital status’, and ‘occupa¬
tion’. The instantiation of an individual person in memory may be activated
by the person’s name or by one or more of the other properties. The advant¬
ages of such a system are that it can successfully handle incomplete or even
incorrect information and can also, because of the connections among units
in the system, lead to ‘spontaneous generalization’ (for example, to the recog¬
nition that people who have professional occupations like ‘lawyer’ and
 Cognitive accounts of second language acquisition 405

‘doctor’, represented by specific units in the system, also tend to be

‘well-educated’, represented by another connected unit).
The main properties of a PDP model are as follows:

1 Processing is carried out in parallel rather than serially. McClelland,

Rumelhart, and Hinton (1986) point out that multiple constraints govern
language processing, with semantic and syntactic factors constantly
interacting without it being possible to say that one set is primary. They
argue that if ‘each word can help constrain the syntactic role, and even the
identity, of every other word’ (1986: 7), processing must take place not
serially but simultaneously on different levels.
2 A PDP model consists of ‘units’ or ‘nodes’ connected in such a way that the
activation of one unit can inhibit or excite others in varying degrees. The
nature of these units varies, depending on what kind of behaviour is being
modelled. In some cases the units stand for hypotheses about, for example,
the syntactic roles of words in a sentence, while in other cases they stand for
goals and actions, and in still others for aspects of things. Units have output
values which they can pass on to other units in the system. Also, in many
PDP models, the units are organized into different ‘levels’, such that any
one unit excites or inhibits other units at both its own level and other levels.
The resulting network is highly complex and is thought to mirror the
neural structure of the brain.
3 Knowledge is viewed in terms of the microstructure rather than macro¬
structure of cognition, i.e. as connection strength rather than as gener¬
alized ‘patterns’ or ‘rules’. As McClelland, Rumelhart, and Hinton put it:
‘In these models, the patterns themselves are not stored. Rather, what is
stored is the connection strengths between units that allow these patterns
to be recreated’ (1986: 31).
4 Processing is activated by ‘input’ which stimulates one or more units and,
of course, their connections with other units. Thus, in computer simula¬
tions of PDP models, selected input in the form of ‘training sets’ is fed into
the computer, often in successive runs. For example, MacWhinney (1989)
tested a PDP model designed to learn German articles by feeding in 102
carefully selected nouns in different case contexts one hundred times.14
5 Learning is a by-product of information processing; as a result of re¬
sponding to input, the association strengths among units are spontan¬
eously modified. Learning consists of discovering the right connection
strengths from the input. Interconnection strengths do not have to be built
into the models by the researcher, but are derived through experience. The
models, therefore, are ‘dynamic, interactive and self-organizing
systems’ (McClelland, Rumelhart, and Hinton 1986: 42).
6 This process of adjusting connection strengths is governed by ‘learning
rules’ such as the Hebb rule, which states: ‘Adjust the strength of the
406 Internal factors

connection between units A and B in proportion to the product of their

simultaneous activation’ (op. cit.: 36). A number of complex learning rules
have been identified.
7 PDP models can go beyond the input by means of ‘spontaneous general¬
ization’. The models can extract ‘regularities’ in the patterns of intercon¬
nections; these resemble higher order rules. However, PDP networks do
not learn ‘rules’; rather they learn to act as though they knew them.
8 PDP models display ‘emergent properties’. That is, they represent learning
as a gradual process involving a series of stages. As we will see, they behave
like language learners in that they produce incorrect as well as target-like
constructions. Importantly, these incorrect responses are generated by the
network itself.

We will now consider Rumelhart and McClelland’s (1986) simulation of

past tense learning using a PDP model. This grammatical feature is of consid¬
erable interest to language researchers because, as we have seen, its natural
acquisition typically involves U-shaped learning behaviour (for example,
went—goed—wented—went). Overgeneralization errors, such as ‘goed’,
have been held to provide strong evidence that learners must organize their
linguistic knowledge into ‘rules’. Given that PDP systems are constructed in
such a way that no rules of this nature are represented, past tense learning
provides a rigorous test of the ability of such systems to account for language
learning.
The PDP system that Rumelhart and McClelland used to simulate past
tense learning is too complex to describe here. In simple terms, it is equipped
with (1) an encoding device that operates on the root form of verbs (for ex¬
ample, ‘hope’) and converts them into a set of context-sensitive phonological
features (called Wickelfeatures); (2) a ‘pattern associator’ network that take
s the output from (1) and computes the past tense form of the verbs, again in
terms of Wickel features; and (3) a decoding device that converts the output
of (2) into a normal set of phonological features. Large numbers of exemplars
of both root and past tense verb forms were fed systematically into the model;
they comprised the input. The pattern associator is also equipped to compare
the output of the network with the target form. When the output matches the
target, the model is doing the right thing and so none of the connection
weights between the units in the pattern associator are adjusted, but when it
does not match, adjustments to the weights are made.
One of the most interesting results obtained by the simulation was that the
pattern observed in the LI acquisition of past tense forms occurred. This was
evident in similar learning curves for different classes of verb, similar stages of
development, and the same errors. For example, the model generated the
familiar U-shaped learning curve, and errors involving double past tense
marking. Thus, for ‘eat’, ‘ated’ occurred later than errors involving over¬
generalization (‘eated’), a phenomenon also reported in LI acquisition
 Cognitive accounts of second language acquisition 407

(Kucczaj 1977). The model not only accounted for the major known features
of acquisition, but also afforded a number of predictions regarding features
not reported in the LI acquisition literature, for example, that overgeneral¬
ization errors are more likely to occur in irregular verbs like ‘come’ whose
past tense form does not involve any modification to the final phoneme of the
base form. Rumelhart and McClelland argue that the simulation demon¬
strated that a ‘reasonable account of the acquisition of past tense can be pro¬
vided without recourse to the notion of “rule” as anything more than a
description of the language’ (1986: 267).
This conclusion is, of course, enormously challenging to mentalist theories
of language learning, as it suggests that learning is environmentally driven
and that learners need very little in the way of innate knowledge. Not surpris¬
ingly, therefore, it has been criticized by researchers working in the Chom¬
skyan paradigm. Pinker and Prince (1989) observe that ‘the fact that a
computer model behaves intelligently without rules does not show that hu¬
mans lack rules, any more than a wind-up mouse shows that real mice lack
motor programs’ (1989: 184). They also point out that Rumelhart and
McClelland’s simulation met with only limited success, as even after extens¬
ive training it achieved only 67 per cent accuracy in a test of the model’s abil¬
ity to generalize its knowledge to new verbs. In many cases, this test resulted
in errors not attested in natural language acquisition (for example, ‘typeded’
for ‘typed’ and ‘membled’ for ‘mailed’). Pinker and Prince suggest that such
errors result precisely because the model has no representation of the ‘root
form of a verb’. They point out that many of the overgeneralizations that chil¬
dren make in LI acquisition cannot be explained with reference to input fre¬
quency.15 For Pinker and Prince, then, connectionist theories are ‘revisionist’
in a way that is ‘not scientifically defensible’ (1989: 198).
Connectionists have responded to these criticisms. They reject the claim
that the theory underlying their work is ‘revisionist’ (i.e. behaviourist) in na¬
ture, pointing out that they are concerned with internal states and mental pro¬
cesses. PDP models do have an ‘initial state’, which is, in a sense, ‘innate’ and,
as Gasser (1990) notes, connectionists are increasingly concerned with what
it consists of. Gasser also points out that there are linguistic theories (for ex¬
ample, Langacker 1987) that are compatible with PDP models in that they
emphasize the continuous nature of the differences between rules and excep¬
tions. Similarly, Schmidt (1988) observes that PDP is good at dealing with
variability and ‘fuzzy concepts’. MacWhinney (1989) also finds that connec-
tionism offers ‘a powerful formal framework that correctly expresses the pro¬
cessing and learning claims of the Competition Model’ (1987: 457). This is
not surprising, perhaps, given that both PDP and the Competition Model
share a number of features in common, such as the notions of ‘network’ and
the ‘weight’ of cues or connections.
Gasser (1990) provides a balanced conclusion:
408 Internal factors

It is now clear that some form of connectionism will figure in a general

model of human linguistic behavior. The only question is whether the role
will be a minor one, relegated to low-level pattern matching tasks and the
learning of exceptional behavior, or whether the connectionist account will
supersede symbolic accounts, rendering them nothing more than approx¬
imations of the actual messy process (1990: 186).
The resolution to this question lies in further research into simulated
language learning, including T2 learning, which to date has received little
attention from PDP researchers.16

Conclusion
The theories that have been examined in this chapter afford a cognitive as
opposed to a mentalist view of L2 acquisition. That is, they seek to explain
how learners acquire an L2 in terms of cognitive processes of a general nature:
‘strategies’, ‘operations’, ‘mappings’, ‘principles’, ‘learning rules’, and the
like. We have made a broad distinction between theories that address how L2
knowledge develops over time and theories that explain learners’ ‘procedural
skill’ (i.e. their ability to use their knowledge in communication). We have
seen, though, that this distinction is not watertight, as many cognitive
theories characterize the way in which L2 knowledge is represented in terms
of how it is actually used. Indeed, it is characteristic of cognitive as opposed to
linguistic theories that this distinction is blurred. In the case of one
theory—Parallel Distributed Processing—it disappears altogether. One of
the major issues in SLA research today is whether L2 knowledge is best
characterized in terms of ‘rules’ which govern the behaviour of specific
items—as claimed by many cognitive theories of L2 acquisition and the
linguistic
theories to be examined in the following chapter—or whether it is best
accounted for in terms of well-practised, discrete items stored either as
self-contained units or as sets of inter-related features, as suggested by some
of the approaches current in cognitive psychology, such as PDP.
It is not yet possible to construct a comprehensive cognitive theory of L2
acquisition, let alone an all-embracing theory that incorporates both a
linguistic and a cognitive perspective. Nor is it easy to evaluate the different
cognitive theories, for they differ in major ways with regard to (1)
provenance, (2) scope, (3) the type of data they aim to account for, and (4) the
precision of the hypotheses they advance. Theories like the Multidimensional
Model have originated in work undertaken by L2 researchers, others were
devised to account for LI acquisition and were subsequently used in SLA
research (for example, Slobin’s operating principles) while still others are
general cognitive theories applied to L2 acquisition (such as the skill-learning
theories of Anderson and McLaughlin). We have seen that some theories
focus on the way input is processed (for example, the Competition Model),
some on the way L2 knowledge is represented (such as the Monitor Theory),
while others
 Cognitive accounts of second language acquisition 409

seek to account for L2 knowledge in relation to output (for example, the Mul¬
tidimensional Model). Some theories have been constructed to account for
naturalistic L2 data (for example, variability theories) while others are based
on data elicited through experiments or simulations (such as the Competition
Model and Parallel Distributed Processing). Finally, there are theories that af¬
ford very precise hypotheses that can be systematically tested (for example,
The Competition Model and the Multidimensional Model) while others offer
more general statements that provide only a broad-brush picture of how ac¬
quisition takes place (for example, theories based on the implicit/explicit
distinction).
In Chapter 15 we will examine a number of criteria which have been used
to evaluate theories, but it is unlikely that any one of the theories we have
looked at here will win out over the others. The plethora of models and theo¬
ries in L2 acquisition has often been noted, and in some cases regretted (for
example, by Schouten 1979). It testifies to two general points. One is that SLA
research is still a very new field of enquiry, scarcely more than twenty-five
years old. The other is the enormous amount of interest which the study of L2
learning has generated.

Notes
1 Various alternative terms to ‘interlanguage’ have been used. Corder
(1971a) refers to ‘idiosyncratic dialects’ and ‘transitional competence’,
whereas Nemser (1971) uses ‘approximative systems’. Although these
terms have different shades of meaning, they all refer to the same general
phenomenon. The term ‘interlanguage’ has stuck and will be the one used
here.
2 Selinker (1972) has made it clear that fossilization applies to both non¬
target and target forms. Thus, it is evident not only when learners produce
errors but also when they produce target-like utterances. In practice,
however, fossilization has been used to refer to ‘persistent errors’ (Muk-
katesh 1986; Major 1988).
3 Krashen’s five hypotheses are: (1) the acquisition-learning hypothesis, (2)
the natural order hypothesis, (3) the input hypothesis, (4) the monitor hy¬
pothesis, and (5) the affective filter hypothesis. In this chapter we focus on
(1) and (4). Hypotheses (2), (3), and (5) are considered in Chapters 3, 6,
and 11 respectively.
4 Schmidt (1990) speculates that parallel distributed processing models
may provide an explanatory basis for implicit learning. That is, learners
build up association strengths rather than ‘rules’ implicitly.
5 Krashen’s position appears to have shifted somewhat over the years, as
subsequent to the publication of Tarone’s 1983 article he has accepted
that variability can occur in unmonitored utterances as a result of what he
410 Internal factors

calls editing by ‘feel’. He talks of ‘inter-stage fluctuation’ (Krashen 1985:

64-5).
6 Tomlin also reports that native speakers conformed to the quantity prin¬
ciple on the same task.
7 The Competition Model can also be seen as a connectionist theory and, as
such, might belong to the section of this chapter dealing with Parallel Dis¬
tributed Processing. However, given the centrality of form-function rela¬
tionships to the model it is appropriate to consider it under ‘functional
theories’.
8 The choice of second noun in unnatural NNV and VNN sentences by nat¬
ive speakers of English has been noted in a number of previous studies and
is considered to reflect the subjects’ knowledge of the transformation
which allows a direct object to take up an initial noun position (as in
‘Mary John likes a lot’) and which, as a consequence, moves the agent
noun into the second noun position.
9 I am indebted to Widdowson for the observation that ‘utterance’ and
‘sentence processing’ do not involve the same processing procedures.
Widdowson also points out that this failure to distinguish utterances and
sentences was one of the reasons why earlier research on language pro¬
cessing based on transformational generative grammar proved to be such
a dead-end.
10 Yoshioka and Doi (1988) report that ‘the subjects who used o correctly at
all also used wa correctly but the opposite direction is not necessarily the
case’ and also that ‘the subjects who used ga correctly even once used wa
and o correctly but the subjects who used wa and o correctly did not ne¬
cessarily use ga correctly’. They point out, however, that a longitudinal
study is needed to test the hypothesis properly.
11 Anderson’s model has changed somewhat over the years. The 1983 ver¬
sion became known as ACT"' (to be read ‘ACT-star’). The distinction be¬
tween declarative and procedural knowledge, however, is basic to all the
ACT models.
12 Faerch and Kasper (1985) also distinguish declarative and procedural
knowledge. However, their definition of the distinction differs from
Anderson’s. For them procedural knowledge consists of procedures for
using declarative knowledge (i.e. procedural skill) and it is not possible to
talk of declarative knowledge transforming into procedural knowledge,
as in Anderson’s model.
13 McFaughlin’s (1990a) account of ‘restructuring’ is not dissimilar to Bia-
lystok’s idea of ‘analysis’. Exemplar-based representation corresponds to
‘unanalysed knowledge’ while rule-based representation corresponds to
‘analysed knowledge’.
14 The selection of ‘training sets’ in PDP models for language is usually
based on information regarding the typical frequency with which the
 Cognitive accounts of second language acquisition 411

specific features or configurations of features that constitute the learning

target occur in natural language use.
15 An additional criticism of PDP accounts of L2 acquisition (not made by
Pinker and Prince) is that it is not clear how a PDP model can produce the
‘restrictive simplification’ evident in all early acquisition. How, for ex¬
ample, can a PDP model produce output consisting of copula-less sen¬
tences given an input that consists entirely of grammatical copula
sentences?
16 Exceptions are Gasser’s own exploratory study of transfer in L2 acquisi¬
tion and Sokolik’s (1990) attempt to illustrate the use of PDP modelling
to explain the difference between child and adult L2 acquisition. It should
also be noted that a language learning simulation, such as Rumelhart and
McClelland’s study of past tense learning, is neutral as to whether it in¬
volves LI or L2 learning. In so far as these authors’ point of comparison
was studies of LI acquisition it can be considered an LI study. However,
their results can also be compared to those obtained from L2 research.
Obviously, though, a PDP model of L2 learning must in some way incorp¬
orate LI knowledge into the initial state.

Further reading
An excellent account of some of the earlier theories considered in this chapter
(Interlanguage Theory; Monitor Theory), and also of his own skill-building
theory, can be found in
B. McLaughlin, Theories of Second Language Acquisition (Edward Arnold,
1987).
D. Larsen-Freeman and M. Long (eds.), An Introduction to Second language
Acquisition Research (Chapter 7) (Longman, 1991) also provides an over¬
view of many of these theories. Their account of the Multidimensional Model
is particularly good.
The following are suggested readings for the individual theories:

1 Interlanguage theory

L. Selinker, M. Swain, and G. Dumas, ‘The interlanguage hypothesis ex¬

tended to children’. Language Learning (1975) 25: 139-91.

2 Theories based on the implicit/explicit distinction

S. Krashen, Second Language Acquisition and Second Language Learning.

(Pergamon, 1981).
R. Schmidt, ‘The role of consciousness in second language learning.’ Applied
Linguistics (1990) 11: 129-56.
412 Internal factors

3 Variability theory

E. Tarone, ‘On the variability of interlanguage systems.’ Applied Linguistics

(1983)4:142-63.
R. Ellis, ‘Sources of variability in interlanguage.’ Applied Linguistics (1985)
6: 118-31.

4 Functional theories

The literature on functional theories tends to be technical and dense. Studies

in Second Language Acquisition, Volume 14, is given over to ‘Grammat-
icalization in second language acquisition’. Various L2 studies based on the
Competition Model can be found in Applied Psycholinguistics, including
Harrington (1987).

5 Operating principles

R. Andersen, ‘Models, processes, principles and strategies: second language

acquisition inside and outside the classroom’ in B. VanPatten and J. Lee
(eds.): Second Language Acquisition - Foreign Language Learning (Multilin¬
gual Matters, 1990).

6 The Multidimensional Model

J. Meisel, H. Clahsen, and M. Pienemann, ‘On determining developmental

stages in natural second language acquisition.’ Studies in Second Language
Acquisition (1981) 3: 109-35.
M. Pienemann and M. Johnston, ‘An acquisition-based procedure for second
language assessment.’ Australian Review of Applied Linguistics (1986) 9:
92-122.

7 Skill-building models

J. Anderson, Cognitive Psychology and its Implications (Freeman, 1980).

B. McLaughlin, T. Rossman, and B. McLeod, ‘Second language learning: an
information-processing perspective’. Language Learning {1983) 33:135-58.

8 Speech planning and temporal variables

H. Dechert, D. Mohle, and M. Raupach (eds.), Second Language Produc¬

tions (Gunter Narr, 1984).
R. Schmidt, ‘Psychological mechanisms underlying second language fluency.’
Studies in Second Language Acquisition (1992) 14: 357-85.

9 Communication strategies

C. Faerch and G. Kasper (eds.), Strategies in Interlanguage Communication

(Longman, 1983).
 Cognitive accounts of second language acquisition 413

E. Bialystok, Communication Strategies: A Psychological Analysis of Second-

Language Use (Blackwell, 1990).

10 Parallel Distributed Processing

R. Schmidt, ‘The potential of parallel distributed processing for S.L.A. theory

and research.’ University of Hawaii Working Papers in ESL (1988) 7: 55-66.
10 Linguistic universals and second
language acquisition

Introduction

As we saw in the previous chapter, explanations of the internal mechanisms

responsible for L2 acquisition can be divided into those that emphasize the
general cognitive nature of acquisition and those that emphasize its specific¬
ally linguistic nature. The former discuss L2 acquisition in terms of ‘pro¬
cesses’, ‘strategies’, and ‘operations’, while the latter refer to ‘linguistic rules’,
‘linguistic principles’, and, increasingly, to ‘linguistic universals’. Whereas
cognitive theories look to psychology, linguistic theories look to linguistics.
It is customary to identify two broad approaches to the study of language,
which, following Chomsky (1986a), can be called the Externalized (E) Ap¬
proach and the Internalized (I) Approach. The former involves the collection
of samples of a particular language or languages, which then serve as the data
for developing a descriptive ‘grammar’. It is concerned with ‘behavior or the
products of behavior’ (Chomsky 1986a: 17), and the task of the linguist is to
describe these. The I-Approach makes use of a native speaker’s intuitions
about what is grammatical and ungrammatical to investigate the abstract
principles that shape the way the grammar of any particular language is rep¬
resented in the mind of a speaker. The object of inquiry is ‘the states of mind/
brain that enter into behavior’ (op. cit.); that is, the knowledge system or
grammar, generally referred to as linguistic competence, that underlies actual
linguistic behaviour.
Chomsky (1988) notes that there has been a move away from an E-
Approach to an I-Approach, as evident in the switch from the structural to the
generative tradition in American linguistics. However, both approaches are
extant in the study of linguistic universals. The study of typological univer¬
sals, initiated by Greenberg (1966), has been continued by Hawkins (1983),
Comrie (1984), and Croft (1990), among others. It involves the crosslinguis-
tic comparison of a wide range of languages drawn from different language
families in order to discover what features they have in common. In contrast,
linguists working in the Chomskyan tradition seek to discover universals by
the in-depth study of individual languages.1 This results in the identification
of the abstract principles that comprise Universal Grammar and that con¬
strain the form of the grammar of any specific language.
416 Internal factors

The universals identified by the two approaches are very different. Typo¬
logical universals are couched in the familiar language of grammatical de¬
scriptions (for example, ‘all languages have nouns and verbs’) and refer to
specific linguistic properties. Universal Grammar consists of principles and
parameters, which are described by means of highly abstract statements relat¬
ing to general properties of language (for example, ‘a language has the heads
on the same side in all its phrases’—the phrase structure principle, which we
will consider later). Such principles may be reflected in a whole series of rules
in a particular language. The two approaches, although procedurally and
conceptually very different, need not be seen as mutually exclusive, however.
It is possible that at least some of the universals uncovered through typolog¬
ical comparison may have a basis in the abstract principles of Universal
Grammar, while many generative linguists now seek to test the validity of the
abstract principles they have uncovered by crosslinguistic comparisons. Also,
many SLA researchers invoke both typological and UG arguments in con¬
structing hypotheses. Nevertheless, it seems safer to examine each separately,
as to do otherwise runs the risk of comparing apples and oranges.
Linguistic universals must be distinguished from other types of universals
(see Selinker 1984). Cognitive theories also make claims about ‘universals’
(for example, processing universals), while, as we have seen, evidence exists
to suggest that there are also developmental ‘universals’ (regularities in the
order and sequence of L2 development). Linguistic universals may influence
and, perhaps, help to explain processing and developmental universals, but
they are distinct from them. Other factors of a cognitive nature, such as per¬
ception, memory, and retrieval, are also involved.

Interlanguage Theory: another perspective

In the last chapter we considered Interlanguage Theory as an example of a
cognitive account of L2 acquisition. However, Interlanguage Theory has also
informed linguistic approaches in SLA research, and so constitutes a useful
starting point for this chapter.
According to Adjemian (1976), the sine qua non of the interlanguage hypo¬
thesis is that interlanguages are ‘natural languages’ and, therefore, subject to
all the same constraints. A number of assumptions follow from this hypo¬
thesis: interlanguages consist of ‘a set of linguistic rules which can generate
novel utterances’ (1976: 299); claims about the structure of interlanguages
can be derived from grammatical theory; and, like natural languages, in¬
terlanguages can be idealized to make them amenable to linguistic analysis.
Adjemian argues that the goal should be to describe and explain the nature of
the learner’s competence at different stages of development by analysing ‘in¬
tuitional data’ collected experimentally; the main focus should be ‘the gram¬
matical nature of a learner’s IL’ rather than ‘strategies’ (1976: 306). In this
 Linguistic universals and second language acquisition 417

way, Adjemian put the case for a linguistic approach to the study of
interlanguage.
The position which Adjemian’s early paper outlined has subsequently been
reaffirmed on numerous occasions. Eckman (1991: 24), for instance, invokes
Adjemian in advancing what he calls the Interlanguage Structural Conform¬
ity Hypothesis, which states:

The universal generalizations that hold for the primary languages also hold
for interlanguages.

Eckman, like Adjemian, claims that interlanguages are languages and, fur¬
ther, that proposed linguistic universals are fully universal, in the sense that
they apply to non-primary as well as primary languages. Eckman provides
evidence from 11 adult Asian learners of English that universal phonological
generalizations (for example, ‘if a language has at least one final consonant
sequence consisting of step + stop, it also has at least one final sequence con¬
sisting of fricative + stop’, 1991: 24) also hold for interlanguages.
The purpose of this chapter is to explore to what extent ‘the IL one is creat¬
ing is susceptible to the force of several types of language universals’ (Selinker
1992: 247). We will consider two broad types: typological universals, and
those found in Universal Grammar.

Typological universals and second language acquisition

Linguistic typology and language universals

A typological approach to linguistic analysis involves a crosslinguistic com¬
parison of specific features such as articles, word order or relative clause con¬
struction. It provides a basis for identifying both which features are rare in a
particular language (for example, the use of inversion after initial adverbials
in English2), and also which features are common. It is the latter that SLA re¬
search has focused on, although, as we saw in Chapter 8, there has also been
some attention given to how L2 learners acquire rare features.
Various kinds of commonalities have been identified. There are a number
of unrestricted or absolute universals, which are exemplified in all languages.
For example, all languages have nouns and verbs and vowels and consonants.
However, there are probably few such universals, universal tendencies being
much more common. A specific feature may be found in a large number of
languages, but be missing from some. For example, Dahl (1979) found that
there is ‘a universal tendency for Neg to have a definite position relative to the
finite element’ (1979: 91). (‘Neg’ here refers to the negator, for example, ‘not’
in English.) There was also a definite preference for the negator to take a pre¬
verbal rather than a postverbal position. As Comrie (1984) acknowledges,
universal tendencies are problematic because they raise the question as to
418 Internal factors

whether it is justifiable to talk of ‘universals’ when there are exceptions and

also just how many exceptions can be tolerated. He argues, however, that the
non-random distribution of alternative features in the worlds’ languages is of
significance.
Cutting across the distinction between absolute universals and universal
tendencies is the distinction between implicational and non-implicational
universals. Frequently, crosslinguistic comparisons demonstrate that there
are connections between two or more features, such that the presence of one
feature implies the presence of another or others. Implicational universals
take the form of ‘if/then’ statements. These can be ‘simple’ or ‘complex’. A
simple implicational universal involves a connection between just two fea¬
tures. For example, Hawkins (1983) provides evidence to show that ‘if a lan¬
guage has a noun before a demonstrative, then it has a noun before a relative
clause’ (1983: 84). A complex implicational universal involves a relationship
between several features. Hawkins gives the following example: ‘If a lan¬
guage is SOV, then if the adjective precedes the noun, then the genitive pre¬
cedes the noun’. Implicational universals have been found to be particularly
prevalent in the area of word order, as these examples show. The important
point about both types of implicational universals is that they logically
preclude at least one combination of features. For example, the example from
Hawkins given above excludes the possibility that languages which have
nouns before relative clauses will also have nouns before demonstrat¬
ives—the implicature of the universal operates in one direction only.
Complex implicational universals sometimes take the form of a universal
hierarchy. A hierarchy covers a chain of implicational universals, as in the
Noun Phrase Accessibility Hierarchy (AH), which we considered in Chapter
3 (see Table 3.7). This concerns the various functions of the pronoun in relat¬
ive clauses. For example, the relative pronoun functions as the subject of its
clause in this sentence:
1 I bought the puppy that made me laugh.
but as direct object of its clause in this sentence:
2 The puppy that I bought was a nuisance.
As we will shortly see, the function a relative pronoun performs in its clause is
separate from the function of the relative clause in the sentence.
Comrie and Keenan’s (1979) crosslinguistic study showed that languages
varied in the noun phrases that were ‘accessible’ to relativization. It proposed
the following hierarchy:

subject < direct object < indirect object < oblique < genitive < object of
comparative

Their study showed that this hierarchy reflected both the frequency of the dif¬
ferent relative pronoun functions possible in the languages they investigated
and also the presence and absence of specific functions in a single language.
 Linguistic universals and second language acquisition 419

For example, any language that permitted the indirect object to be relativized
would also allow relativization of all the other noun phrase functions above it
in the hierarchy (i.e. direct object and subject), but not below it (i.e. oblique,
genitive, and object of comparative). The hierarchy was also found to be rel¬
evant for explaining the presence or absence of pronominal copies in a lan¬
guage. Thus, a pronominal copy (for example, ‘*The puppy that I bought it
was a nuisance.’) was most likely to occur with relative pronouns lower down
the hierarchy than with those higher up.
The AH, as originally proposed by Comrie and Keenan, is problematic in a
number of ways. Keenan (1975) points out that indirect object and oblique
are indistinguishable in most languages (including English), as both are ex¬
pressed as prepositional phrases, for example:

The official to whom Mary gave the present is sick.

The official to whom Mary spoke is sick.

Jones (1991a) argues that it is a mistake to include genitive in the hierarchy, as

it has a separate and complete hierarchy of its own. He suggests that there are,
in fact, two hierarchies, one for -Gen and the other for +Gen, as shown in
Table 10.1. As we will see, this proposal is given support by the very mixed
empirical results obtained for genitive, when this is viewed as part of a single
hierarchy. Finally, object of comparison is problematic in English—a lan¬
guage that is supposed to manifest this function—because many native
speakers do not accept that sentences with this function are grammatical.

Function - Genitive + Genitive

Subject (SU)
Direct object (DO)
Indirect object (10)
Object of preposition (OP)
Object of comparative (OC)
The man who came ... The man whose wife
came...
The man (whom) I saw ... The man whose wife I saw ...
The man (whom) I gave the The man whose wife I gave
book to... the book to...
The man (whom) I looked The man whose wife I looked
at... at...
The man (whom) I am bigger The man whose wife I am
than ... bigger than...

Table 10.1: The Accessibility Hierarchy for-Genitive and +Genitive (from Jones
1991a)

Relativization is not the only problem that learners face in acquiring relat¬
ive clauses. There is also the question of where the relative clause occurs in the
matrix sentence. Relative clauses can function as part of the direct object of
the sentence, as in (1) above, or they can function as part of the subject of the
sentence, as in (2). In (1) the relative clause is joined onto the main clause,
while in (2) it is embedded in the main clause. As we saw in Chapter 3,
420 Internal factors

whether the relative clause is joined onto or embedded in the main clause can
affect L2 acquisition.
Of considerable importance for SLA research is the way markedness is
handled in language typology. Whereas classical markedness, as defined by
linguists in the Prague School of Linguistic Theory (for example, Trubetzkoy
1931), views features as either unmarked or marked (for example, in the pair
of features ‘a’ and ‘an’, ‘a’ is unmarked and ‘an' is marked), language typo¬
logy sees markedness primarily as a relative phenomenon (i.e. one feature is
more marked than another). Implicational universals presuppose a marked¬
ness relationship; if a language that has property X also has property Y, then
Y is unmarked in relation to X. Hierarchies reflect degrees of markedness
even more clearly. Thus, in the AH, the least marked function is that of sub¬
ject and the most marked is that of object of comparative.
What, then, determines the level of markedness of specific linguistic
features? Croft (1990) reviews Greenberg's (1966) early work on linguistic
universals and identifies three main types of evidence:''

1 Structure: this concerns the presence or absence of a feature. For example,

plural can be considered more marked than singular because it typically
involves the addition of a morpheme.
2 Behaviour: this concerns whether one element is grammatically more
‘versatile’ than another—the more versatile it is, the more unmarked it is.
Versatility is evident in both the number of inflections a specific grammat¬
ical category possesses (for example, singular 3rd person has three forms in
English—‘he’, ‘she’, and ‘it’—whereas plural third person has only
one—‘they’) and in the number of syntactic contexts in which a specific
grammatical element can occur (for example, more constructions occur
with the active voice than with the passive voice).
3 Frequency: the unmarked value is likely to occur with greater frequency
than the marked value, both in actual use (i.e. in actual texts) and also in
the world’s languages.

Croft considers that ‘markedness is a fundamental concept, underlying much

grammatical work in typology’ (1990: 64).
To summarize, linguistic typology involves the crosslinguistic study of
samples of the world’s languages. It results in the identification of various
types of universals: unrestricted universals, universal tendencies, implica¬
tional universals, and universal hierarchies. Also, it provides a principled
basis for determining the degree of relative markedness of connected features.

Typologically motivated studies of second language acquisition

SLA research has used linguistic typology to seek answers to three main
questions:
 Linguistic universals and second language acquisition 421

1 What influence do typological universals have on the order of acquisition

of grammatical features?
2 What effect does markedness have on learning difficulty?
3 How does the typological status of grammatical features in the native and
target languages affect LI transfer?

We have already considered (3) in Chapter 8 (see pages 319-24), so we will

focus our attention on (1) and (2) here.
The research based on linguistic typology has examined phonological fea¬
tures (for example, Eckman 1977,1984, and 1991; Tarone 1980a) as well as
grammatical features (for example, Levenston 1971; Hyltenstam 1977;
Schmidt 1980; Rutherford 1983; Zobl 1989; Jones 1991b). No attempt will
be made to provide a comprehensive review of this literature. Instead, se¬
lected studies will be considered as a way of examining the major theoretical
issues.

Negative placement

Negative placement constitutes an example of a non-implicational universal

tendency, with Dahl’s (1979) survey showing a clear preference for preverbal
over postverbal negative position in the world’s languages. As we saw in
Chapter 3, there is strong evidence that in the early stages of L2 acquisition
learners opt for preverbal negation, even where the LI manifests postverbal
negation. If, on the basis of language typology, preverbal negation is con¬
sidered unmarked in relation to postverbal negation, it would seem that
learners acquire the unmarked form before the marked. Furthermore, as
Wode (1984) notes, preverbal negation appears even when it is not present in
either the target language or the learner’s LI. This suggests that typological
universals like negator + verb phrase may occur irrespective of the formal
characteristics of the languages involved.
Jordens’ (1980) re-analysis of Hyltenstam’s (1977) data on the acquisition
of the negative placement rule in Swedish by Turkish, Serbo-Croat, and
Hamito-Semitic speakers lends support to a number of predictions based on
typological markedness. Swedish is of particular interest because ‘in main
clauses the negation is placed after the finite verb, while in subordinate
clauses it is placed immediately before the finite verb’ (Hyltenstam 1977:
387). As such, it appears to contradict the ‘universal tendency for Neg to have
a definite position relative to the FE (finite element) of the sentence’ (Dahl op.
cit.: 91). Jordens argues that Hyltenstam’s learners initially went for a nega¬
tor + main verb ordering in both main and subordinate clauses:

*han inte komme (he not comes)

att han inte kommer (that he not comes)

The learners, therefore, appear to be behaving like other learners by opting

initially for a negator in preverbal position (as reported in many studies of L2
422 Internal factors

negation) and also by placing the negator in a definite position. Jordens also
points out that this position is initially determined not by the finite verb ele¬
ment (as this is not yet represented in the learners’ interlanguage) but, more
simply, by the main verb.4

Preposition stranding and pied piping

Preposition stranding and pied piping have been investigated by Mazurke-
wich (1984) and Bardovi-Harlig (1987). Preposition stranding is typo-
logically rare and, therefore, might be expected to be acquired late. It is
considered marked in relation to pied piping (see van Riemsdijk 1978 cited in
Bardovi-Harlig 1987) on the grounds that any language that has preposition
stranding also has pied piping, but not the reverse. Also, the relationship is
considered to be implicational (i.e. languages that have preposition stranding
also have pied piping). Both structures are found in w/i-questions and relative
clauses in English, as shown in these examples:

With whom did Mary speak? (pied piping)

The man with whom Mary spoke ... (pied piping)
Who(m) did Mary speak with? (preposition stranding)
The man who(m) Mary spoke with ... (preposition stranding).

Mazurkewich used a written question formation test to investigate the ease of

acquisition of the two structures in ^-questions by French and Inuktitut
speakers. The results indicated that the French speakers produced more in¬
stances of pied piping than preposition stranding, while the opposite was true
for the Inuktitut speakers. In other words, the results were ambivalent, not
lending clear support to the hypothesis that the unmarked structure would
prove easier than the marked. One possible explanation for this could be FI
transfer; Inuktitut does not have prepositions and so, arguably, transfer is not
an issue, whereas French has pied piping but not preposition stranding. The
French learners may have transferred their FI structure, therefore. However,
this is not Mazurkewich’s preferred explanation. She claims that the Inuktitut
speakers were more advanced than the French speakers and, thus, more likely
to have acquired the marked preposition stranding. She also argues that FI
transfer is not an adequate explanation of the French learners’ results because
the Quebec dialect, which they spoke, does manifest incidences of preposition
stranding (for example, le boss que je travaille pour = the boss that I work
for). Mazurkewich’s arguments are dismissed by both Kellerman (1985b)
and White (1986), both of whom consider that the transfer explanation is
stronger.
Bardovi-Harlig’s (1987) study indicates that the markedness hypothesis is,
in fact, not tenable for pied piping/preposition stranding. This study used a
similar linguistic manipulation task to Mazurkewich, but included items
involving relative clauses as well as ^-questions. The subjects were 95
 Linguistic universals and second language acquisition 423

learners of English, divided into six proficiency levels. They had a variety of
Lis, but in every case the LI allowed only the unmarked pied piping. The res¬
ults provided clear evidence that learners acquire preposition stranding be¬
fore pied piping, contrary to the markedness hypothesis. Bardovi-Harlig also
found that before learners attempt either rule, they omit the preposition en¬
tirely (a ‘No-Prep strategy’), as in these examples:

Who did Mary give a book?

The man Mary baked a cake was Joe.

The order of acquisition, therefore, was (1) No-Prep, (2) preposition strand¬
ing? (3) pied piping. This order was evident in both ^-questions and relative
clauses, although relative clauses, as the more difficult structure, were more
likely to retain the ‘simpler’ forms (No-Prep and preposition stranding) than
^-questions. Bardovi-Harlig argues that the results indicate that the learn¬
ers were not responding to linguistic factors (i.e. markedness) but to the avail¬
ability of data in the input (i.e. salience). Thus, preposition stranding is
acquired before pied piping because it is more frequent in the input. She sug¬
gests the following restatement of the markedness hypothesis: ‘unmarked
structures are acquired before marked structures, all things being equal’
(1984:402).

Relativization

Pour major questions have figured in the research based on the Accessibility
Hierarchy (AH): (1) does the AH account for avoidance behaviour? (2) does
the AH explain the order of acquisition of relativizable NP positions? (3) does
the AH explain the use of pronominal copies? and (4) does the AH explain the
acquisition of the forms of different relativizers? In each case, the central issue
is whether markedness factors, typologically determined, can be used to ex¬
plain acquisitional phenomena.
Schachter’s (1974) study of relative clauses, which we considered in Chap¬
ter 8, showed that learners of English whose LI does not contain relative
clauses or contains left-branching rather than right-branching clauses (for ex¬
ample, Japanese and Chinese) tend to avoid using relative clauses in their
English production. Schachter did not investigate whether the extent of this
avoidance was related to the function of the relative pronoun. However, Gass
(1980: 138) did so. She used a sentence-joining task to elicit use of all the rel¬
ative pronoun functions in the AH. She reported that her subjects (adult
learners with mixed LI backgrounds) tended to avoid relativizing on low
positions in the AH by changing a part of one of the sentences in a sentence¬
joining task so as to make relativization in a higher position possible. For
example the two sentences:

He saw the woman.

The man kissed the woman.
424 Internal factors

were intended to elicit the use of a relative pronoun functioning as direct

object:

He saw the woman that the man kissed.

However, several learners contrived to use a subject relative pronoun, as in:

He saw the woman who was kissed by the man.

Akagawa (1990), however, found no support for the hypothesis that Jap¬
anese learners display more avoidance on lower positions than on higher po¬
sitions in the AH. Thus, although there is clear evidence that differences
between the LI and L2 structure of relative clauses induce avoidance, to date
there is only mixed evidence to show that the degree of avoidance corres¬
ponds to markedness.
There is much clearer evidence that markedness, as defined by the AH, in¬
fluences the order of acquisition. Gass (1980) found strong support for the
markedness hypothesis. Thus, fewest errors were evident in the subject posi¬
tion, with the most errors in the object of comparison position. However,
Gass reported that the learners found the genitive function easier than the dir¬
ect object function, contrary to predictions based on the AH, a point that will
be commented on below.
Pavesi (1986), in a comparative study of instructed and naturalistic Italian
learners of L2 English that used implicational scaling on elicited oral data,
found that, in general, both groups of learners followed the order predicted
by the AH. However, the pattern for indirect object and oblique and also for
genitive and object of comparative was not always as expected. Learners var¬
ied with regard to which function in each pair they favoured, some preferring
the unmarked before the marked member and others vice versa. Also, the gen¬
itive and object of comparison functions were almost entirely missing in the
naturalistic group.
Further evidence in support of the markedness hypothesis can be found in
Eckman, Bell, and Nelson (1988) for learners with mixed Lis, and Jones
(1991b) for Japanese college students. In both studies, the accuracy order for
the different pronoun functions on sentence-joining tasks conformed broadly
to the AH, although, again, there were some discrepancies. Jones’ study is of
interest because he provides separate accuracy orders for genitive and non¬
genitive structures. He found that the subject function was considerably
easier than the direct object and oblique positions (which were roughly equal)
irrespective of whether the structure involved a genitive or not. Also, al¬
though non-genitive structures proved easier overall than genitive structures,
accuracy levels on the genitive subject position were equivalent to those on
non-genitive direct object and oblique positions. The importance of consider¬
ing genitive and non-genitive structures separately is borne out by Hansen-
Strain and Strain (1989), who found that the results for genitive in their study
involving five different LI groups on seven different tasks did not match the
 Linguistic universals and second language acquisition 425

AH. The learners performed much better than expected, possibly because
they were only required to produce genitive subject structures.
Three studies have examined pronominal copies in relative clauses in terms
of the AH. The studies by Gass (1979; 1980) and Hyltenstam (1984) have al¬
ready been considered in Chapter 8. The conclusion was that markedness in¬
teracts with LI transfer in the sense that although pronominal copy errors
were more likely in the lower positions and less likely in the higher positions
of the AH, the overall error frequency reflected whether or not pronominal
copies occurred in the learners’ LI. Gass felt that ‘it is universal principles that
play the leading role since they are dominant in assigning a relative order of
difficulty’ (1980: 140). A third study—Tarallo and Myhill (1983)—casts
doubt on this conclusion, however. English LI subjects were asked to judge
the grammaticality of sentences containing pronominal copies in two right¬
branching languages (German and Portuguese) and two left-branching (Chi¬
nese and Japanese). Tarallo and Myhill reported that the learners of right¬
branching languages were most accurate in judgements of sentences involv¬
ing the subject function, while those of left-branching languages were most
accurate in judgements involving the direct object function. In other words,
the learners of the right-branching languages conformed to the order pre¬
dicted by the AH but those of left-branching languages did not. Tarallo and
Myhill concluded that the crucial factor was the proximity of the relativized
noun phrase (NP) site in the embedded sentence to the head of the relative
clause. Thus, in English, the direct object function is more difficult than the
subject function because the extraction site (shown by a_) is further from
the head NP (in italics) than is the case for subject function, for example:

The puppy that I bought_was a nuisance.

I bought the puppy that_made me laugh.

The opposite is the case in left-branching languages.

Finally, Hawkins (1989) examined the acquisition of the different forms of
L2 French relativizers by three groups of adult English-speaking learners,
who differed in their general level of proficiency. He found that the order of
acquisition of the three relativizers, qui, que, and dont, conformed to the or¬
der predicted by the AH, but argued that this order could also be predicted on
the basis of surface configurational factors. For example, qui could be ex¬
pected to be acquired before que, which in turn would be acquired before
dont because this order corresponds to the distance between the head and the
extraction site, as shown in the following examples:

L’homme qui_connait Pierre ... (The man who knows Pierre).

L’homme que Pierre connait___ (The man who Pierre knew).
Le visiteur dont j ’avais oublie le nom-(The visitor whose name I had
forgotten).
426 Internal factors

It is not possible, therefore, on the basis of these results, to decide between a

‘relational view’ (based on the AH) and ‘a configurational view’ (based on the
surface relationships between grammatical elements). However, Hawkins
went on to provide evidence to suggest that the configurational view is the
right one, arguing, for example, that only this view can explain why the learn¬
ers experienced greater difficulty with stylistically inverted relative clauses
like:

L’homme que connait Pierre ... (The man that Pierre knows ...)

than with relative clauses displaying the more conventional order. Such a
conclusion may be premature, however, given that the difficulty with such
stylistically inverted sentences may derive from Ll-related factors (English
does not permit subject-object inversion in relative clauses) rather than from
configurational factors per se.
It is probably premature to reach firm conclusions regarding the effect of
markedness, as represented by the accessibility hierarchy, on L2 acquisition.
There is some evidence to show that linguistic markedness may have an effect
on the extent to which learners avoid relative clauses, the order in which they
acquire relative pronoun functions, and the extent to which learners make
pronominal copy errors. The general finding is that acquisition is easier in the
unmarked, higher positions and more difficult in the marked, lower positions
of the hierarchy. There is, however, an alternative view advanced by Tarallo
and Myhill (1983) and Hawkins (1989), namely that learners construct rules
for relative clauses on the basis of the adjacency of categories in the surface
configuration. According to this view, the difficulty that learners experience
in learning relative clauses is not the product of the relative markedness of the
relative pronoun functions, but a function of their capacity for processing
sentences. We should also note that the L2 research suggests that there are
problems with the AH, as formulated by Keenan and Comrie (1977). In par¬
ticular, the fact that variable results have been found for genitive suggests that
Jones’ (1991a) proposal that there are in fact two hierarchies, one for genitive
and the other for non-genitive, is right.

The role of typological universals in second language acquisition

What conclusions can we reach regarding the role of typological universals in
L2 acquisition? The evidence from the studies of negation, preposition
stranding/pied piping, and relativization which we have considered, indicates
that the linguistic nature of the target structures may influence both the ease
and the order of their acquisition. Learners seem to find it easier to acquire
typologically unmarked structures than typologically marked structures. The
assumption that underlies this research, namely that linguistic universals will
affect interlanguages in the same way as they affect other languages, because
ILs constitute natural languages, is supported.
 Linguistic universals and second language acquisition 427

It is clear, however, that typological markedness alone cannot account for

L2 acquisition. Gass (1984) suggests that it constitutes not so much an ‘abso¬
lute constraint’ as an ‘overall shaping factor’. Bardovi-Harlig’s (1987) study
of preposition stranding/pied piping shows that in some features at least, ac¬
quisition may not proceed in accordance with markedness. If a linguistically
marked feature is strongly represented in the input learners are exposed to,
then they will acquire this before a related unmarked feature that is only
weakly represented. In other words, input frequency can over-ride the as¬
sumed difficulty of learning a marked feature. This raises a thorny question. Is
it markedness per se or input frequency that is the real determinant of acquisi¬
tion? In general, we can assume that unmarked features are likely to be fre¬
quent. Subject relative pronouns, for instance, occur more frequently than
objective of comparison pronouns, and probably also more frequently than
indirect object pronouns. The fact that learners acquire subject relative pro¬
nouns before object of comparison and indirect pronouns may therefore
simply reflect input frequency. In order to decide whether it is markedness or
input frequency that determines order of acquisition, it is therefore necessary
to look at cases like preposition stranding/pied piping in English, where the
two factors do not correlate. Bardovi-Harlig’s finding that learners acquire
the marked but frequent structure first is very damaging to the markedness
hypothesis.
It would be premature, however, to dismiss markedness. Learners do not
always respond to what is in the input, as the studies of negation show. The
input shows learners of L2 English that negation in English follows the auxili¬
ary verb and precedes the main verb, and yet learners almost invariably opt
for preverbal negation in the initial stages. In this case, therefore, they appear
to choose the unmarked form rather than the form supplied by the input.
Again, however, the question arises as to whether the initial preference that
learners show for preverbal negation reflects markedness or some other fac¬
tor. For example, if learners ‘simplify’ the input by ignoring auxiliaries, Eng¬
lish negation will appear to be preverbal. An explanation of early preverbal
negation based on ease of processing is as likely, if not more so, than one
based on typological markedness. According to the Multidimensional Model
(see Chapter 9), for instance, preverbal negation occurs because learners have
not yet removed the processing ‘block’ that prevents them inserting an
element (the negator) into the middle of a string.
The underlying problem is to explain how typological universals affect ac¬
quisition. Two types of explanation are possible. One is that they affect learn¬
ing directly because the learner has inbuilt grammatical knowledge that
includes a theory of grammatical markedness. This is tantamount to sug¬
gesting that learners bring a knowledge of markedness relations, such as
those represented in the AH, to the learning task, and thus are predisposed to
expect unmarked grammatical features to be more likely than marked ones.
This is the view that Hawkins (1989) considers and rejects, probably rightly.
428 Internal factors

The alternative view is that markedness relations are only indirectly related to
language acquisition, and that to understand how they work it is necessary to
uncover the factors that cause one linguistic feature to be more marked than
another. This is the view taken by most L2 researchers. For example, Fiylten-
stam (1984) is careful to argue that the AF1 can only serve as a basis for mak¬
ing predictions about L2 acquisition and does not, in itself, provide
explanations. There are reasons why some features are more common in the
world’s languages than others and it is these reasons that will explain
acquisition.
What then are the factors that determine markedness? Gass and Ard
(1984) identify a number of potential sources of language universals, which
are summarized in Table 10.2. They acknowledge that their list is not ex¬
haustive, and also that any one universal may have more than one source.
‘There is a great deal of controversy with regard to the assignment of an origin
to a particular universal’ (1984: 35).

Source Description

1 Physical basis The universal reflects ‘a physical fact, perhaps dependent on the
way the world is, or... on the way the human body, especially the
vocal and aural apparatus, is structured’.
2 Perceptual/ The universal reflects ‘factors in the human perceptual and cognitive
cognitive apparatus and processing capabilities’. Such factors affect more
than just language.
3 Language The universal reflects innate knowledge of language that a learner
acquisition brings to the learning task.
device
4 Neurological The universal reflects neurological predispositions towards the use
basis of language of certain types of structure.
acquisition
5 Historical change Certain patterns may be missing from all languages because they
cannot diachronically arise from the types of language extant.
6 Social interaction Universals of interactive competence influence the universal nature
of language.

Table 10.2: Sources of linguistic universals (from Gass and Ard 1984: 35-38)

While Gass and Ard’s attempt to uncover the factors that underlie universals
is admirable, it does not take us very far in explaining acquisition. As they ad¬
mit, their framework is programmatic.
Another difficulty of the typological perspective is that it places undue em¬
phasis on syntactic, surface-level features. Rutherford (1982) has argued that
linguistic universals should not be considered in isolation from discourse
function, pointing out that ‘one cannot hope to formulate meaningful gener¬
alizations about syntactic/discoursal order of acquisition by studying syntax
alone’(1982: 103).
 Linguistic universals and second language acquisition 429

Gregg’s (1989) verdict on typological universals is:

... typological generalizations are either uninteresting (no language has

labiovelar stops) or interesting only in the questions they raise, not in the
answers they provide (1989: 32).

Rutherford (1984b: 142) also feels that typological universals are of limited
value—they offer little more than ‘a collection of observations’, albeit ‘con¬
cerning a rather wide assortment of syntactic phenomena’. In so far as typo¬
logical universals explain nothing, Gregg and Rutherford are right. Until it is
possible to determine the precise source of specific universals, the typological
approach will not be capable of explaining L2 acquisition. Typological uni¬
versals still have a place in SLA research, however. They provide a basis for
forming predictions about acquisition and are of value in what Hyltenstam
refers to as ‘the descriptive phase of research’ (1990: 33).
Theorists like Gregg who dismiss typological universals out of hand are
therefore overly harsh. However, an approach that concentrates on the na¬
ture of the abstract knowledge that determines the structure of any particular
language is likely to be more revealing, for, as Lightbown and White (1987)
point out, knowledge of universals and markedness might follow from this
abstract knowledge and the properties of the human mind that give rise to it.
We now turn to the Chomskyan view of linguistic universals and examine in
what ways and to what extent this affords an explanation of L2 acquisition.

Universal Grammar and language acquisition

Whereas language typology is essentially data-driven and descriptive in na¬
ture, the approach that we will consider in this section is motivated by a
powerful theory of language and a well-developed model of grammar. The
theory is Universal Grammar (UG) and the model Government/Binding, as
proposed by Chomsky (1976; 1981a; 1981b). We will focus attention on the
theory. No attempt will be made to summarize the Government/Binding
Model, although aspects of it will be considered by way of illustration. ’

The Theory of Universal Grammar

One of the claims of L2 theorists working within a generative grammar
framework is that any theory of L2 acquisition that is not based on an ad¬
equate theory of language will prove inadequate (Gregg 1989). The argument
is that the Theory of Universal Grammar, as proposed by Chomsky, consti¬
tutes the best theory of grammar currently available, because it achieves both
descriptive and explanatory adequacy. It follows that L2 acquisition research
should be informed by this theory.
The terms ‘descriptive’ and ‘explanatory adequacy’ refer to two key ideas
in the study of grammar. Speakers of a language manifest a ‘steady state of
430 Internal factors

knowledge, which changes only in relatively insignificant ways (such as in vo¬

cabulary). To achieve descriptive adequacy, the linguist must provide a com¬
plete and explicit account of the steady state. To achieve explanatory
adequacy, the linguist must account for how speakers arrive at the steady
state from the initial state, given the nature of the linguistic data they are ex¬
posed to; that is, they must explain how children come to know their mother
tongue. The Theory of UG, therefore, is both a description of language (more
specifically, of grammar) and a theory of how knowledge of a language is
acquired.
This section will address three questions:

1 What does UG consist of?

2 What role does UG play in LI acquisition?
3 What should the domain of a theory of L2 acquisition be?

Readers should note that Chomsky himself has been concerned only with (1)
and (2), and has had almost nothing to say about L2 acquisition.

Principles, parameters, and markedness

Chomsky defines UG as ‘the system of principles, conditions and rules that
are elements or properties of all human languages’ (1976: 29). In other
words, it comprises a set of linguistic universals. Subsequently, Chomsky
(1981a) characterizes these universals as consisting of principles and para¬
meters. The term ‘principles’ refers to highly abstract properties of grammar
which apply to language in general and which, therefore, underlie the gram¬
matical rules of all specific languages. Although the full range of principles
will not be evident in all languages, there will be no language that contravenes
any principle. The term ‘parameters’ refers to principles that vary in certain
restricted ways from one language to another. That is, they take the form of a
finite set of options which individual languages draw on and which define the
variation possible between languages. Parameters have two or more ‘set¬
tings’, with different languages manifesting different settings. Chomsky
(1988) likens parameters to the array of switches found in a switch box; the
learner’s task is to use experience to determine which position each switch
must be in. The goal of generative grammar is to identify the principles and
parameters that comprise UG and to specify which principles and which
parameters are operative in specific languages.
An example of a principle is subjacency. This defines the restrictions that
govern how far one phrase can be moved from ‘deep’ to ‘surface’ structure.
Thus, questions like:

1 What did Randy think?

2 What did Randy think his brother had won?
 Linguistic universals and second language acquisition 431

are grammatical because they involve limited movement of the wh-element

(i.e. ‘what’) from the deep structure object position:

Randy thought_
Randy thought his brother had won_

Sentences like the following, however, are ungrammatical:

3 *What did Randy wonder whether his brother would win?

because they involve movement of the wh-e lement from a ‘remote’ deep
structure position:

Randy wondered whether his brother would win_

In the Government/Binding Model, movement is discussed very precisely in

terms of the restrictions imposed by ‘bounding nodes’, which for English con¬
sist of noun phrase, embedded sentence, and sentence. According to Berwick
and Weinberg (1984) ‘movement may not cross more than one bounding
node’ at a time. Thus, sentences (1) and (2) above are grammatical because
the wh-element does not cross more than one bounding node, whereas (3) is
ungrammatical because it crosses two embedded sentence nodes. It should be
noted, however, that what constitutes a bounding node is not itself universal
and that languages vary in the extent to which movement is restricted. This is
a point that will be illustrated later when we consider a number of L2 studies
based on the subjacency principle.
An example of a parameter is ‘pro-drop’. Languages vary according to
whether they forbid the deletion of subject pronouns. English, for example,
does not normally delete pronouns:6

*Is the President of the United States.

Spanish, on the other hand, permits null pronoun subjects:

Esta el Presidente de los Estados Unidos.

The pro-drop parameter, therefore, ‘determines whether the subject of a

clause can be suppressed’ (Chomsky 1988: 64). It has just two settings (other
parameters may have multiple settings).
Parameters like pro-drop are of considerable interest to linguists, and, as
we will see, also to SLA researchers, because they involve a number of lin¬
guistic features. Chomsky (1981a) has proposed that languages with null
subjects, like Spanish and Italian, do not have expletives (for example,
dummy ‘it’ and ‘there’ in English) and also permit variable word order, with
the verb appearing both before and after the subject in sentences like:

Viene la chica (Is coming the girl).

La chica viene (The girl is coming).
432 Internal factors

In contrast, non-pro-drop languages like English have expletives and also

manifest a fixed word order with the subject placed invariably before the verb
in declarative sentences:

*Is coming the girl.

The girl is coming.

In addition Chomsky suggests that ‘that’-trace effects occur in pro-drop lan¬

guages but not in non-pro-drop:

Chai hai detto che e venuto?

*Who have you said that is come?
(Examples from Gass 1989: 509)

Hyams (1983) has also claimed that non-pro-drop languages have a class of
modal verbs, distinct from main verbs. Finally, Jaeggli and Hyams (1988)
have proposed that null subjects only occur in languages with a uniform pat¬
tern of inflections (i.e. either all verbs are inflected, as in Spanish, or none, as
in Japanese). These various grammatical properties—all entailed by the pro¬
drop parameter—constitute a cluster.7
UG also provides a basis for determining markedness. The degree of mar¬
kedness depends on whether a feature is part of the ‘core’ or the ‘periphery’.
These notions have already been introduced in Chapter 8 (see Figure 8.1).
The core features of a language are those that are governed by UG, while peri¬
pheral features are those that are not. Core features are considered unmarked
because they require minimal evidence for acquisition, whereas peripheral
features are considered marked because they require much more substantial
evidence. The degree of markedness of a feature can also vary within the core,
depending on the parameter setting involved. Parameter settings can be or¬
dered according to how marked they are. Thus, for example, Hyams con¬
siders pro-drop to be unmarked in relation to non-pro-drop, with Spanish
and Italian unmarked with respect to this parameter, and English marked. As
White (1989a) points out, this view of markedness differs from that found in
language typology because markedness is seen as internal to the learner, a
consequence of the language faculty, rather than as something external, evid¬
ent only in extant languages.
It is also possible to identify another rather different definition of marked¬
ness based on UG theory. Zobl (1983b) advances a notion of markedness
based on the learner’s projection capacity, as shown in Figure 10.1. As in the
Theory of UG, markedness is understood in relation to the amount of prim¬
ary linguistic evidence needed to acquire a given property. Zobl’s Projection
Model considers property z unmarked in relation to v, w, x, and y, on the
grounds that the acquisition device does not require any actual experience of
z in order to acquire it (i.e. z does not have to be attested in the input to which
the learner is exposed). Instead, learners are able to infer the existence of z
once they have discovered that certain other properties exist. Clusters of
 Linguistic universals and second language acquisition 433

features such as those associated with the pro-drop parameter—provide a

basis for projection; evidence of one feature in a cluster may enable learners to
acquire the other features associated with it, irrespective of whether they have
experienced these features in the input. Markedness based on a projection
model is also a learner-internal phenomenon.

v
w
— ^ + projective
-►
y capacity
X

Figure 10.1: A projection model of markedness (Zobl 1983b: 294)

Universal Grammar and first language acquisition

The Theory of Universal Grammar is intimately tied to a theory of LI acquisi¬

tion. Chomsky has consistently argued that UG principles are inherently im¬
possible to learn, and that therefore, they must be innate. They make up the
‘initial state’, and as such provide the basis that enables a child to acquire a
language. A child has access to a language acquisition device (LAD) that maps
experience into the ‘steady state’, as shown in Figure 10.2.

experience LAD >- grammar

Figure 10.2: Chomsky’s model of language acquisition (Chomsky 1981a)

Two key issues arise out of this model: (1) what is the nature of the child’s
‘experience’ of the target language, and (2) what does the LAD consist of?
The first question leads to a discussion of what has become known as the
poverty of the stimulus, while the second involves an account of the language
faculty.

The poverty of the stimulus

Generative theorists consider the experience which the young child has of the
target language to be seriously impoverished in a number of ways. It was ini¬
tially argued that input is degenerate (see Miller and Chomsky 1963; Chom¬
sky 1965; McNeill 1966), in the sense that it contains ungrammaticalities and
disfluencies which make it an inadequate source of information for language
acquisition. The principal argument was that children would find it imposs¬
ible to distinguish between what is grammatical and what is ungrammatical
on the basis of such input. Subsequent research into caretaker talk, however,
has demonstrated that simplified input of the kind most children experience
in the early stages of acquisition is far less degenerate than was claimed (see
Chapter 6).
The stimulus might be considered ‘degenerate’ in another way. Wexler and
Culicover (1980) argued that the problem lies not in ungrammaticality but in
the fact that the input is simplified, as this deprives children of the data they
434 Internal factors

require to learn the more complex aspects of grammar. This view has since
been endorsed by White (1989a) and Sharwood Smith (1986), among others.
There are problems with this argument. First, the fact that the input is ‘simpli¬
fied’ does not mean that children are deprived of data on all the grammatical
features they have not yet learnt. Even simplified input may supply the child
with some new information. Second, the caretaker research shows that input
is only temporarily simplified; as the child’s ability to understand grows, so
the input becomes progressively more complex (see Wells 1985). The child,
then, receives ‘full’ input in the course of time. The argument that children do
not receive adequate input for acquisition because it is simplified does not
seem to hold.
Flowever, there are other more compelling reasons than ‘degeneracy’ for
considering the input impoverished. Input seriously underdetermines the fi¬
nal grammar. The child will only be exposed to a subset of the total sentences
possible in the target language and has no way of determining whether a given
sentence is not heard because of coincidence (i.e. it just happens that the input
to date has not provided evidence of it) or because it is not possible in the lan¬
guage. Furthermore, the input does not provide the child with the data needed
to determine that certain constructions are not possible. White (1981) gives
dative alternation as an example. English permits two constructions with
many dative verbs, as in these examples:

Randy gave a present to Mary, (noun phrase (NP) + prepositional phrase

(PP))
Randy gave Mary a present. (NP + NP)

However, other dative verbs permit only the NP + PP pattern:

Randy explained the problem to Mary.

* Randy explained Mary the problem.

Supposing the child works out that many verbs allow both patterns, how can
the restriction on verbs like ‘explain’ be discovered? Again, positive evidence
will not suffice, as the child has no way of knowing that, in time, sentences
with NP + NP will not occur. White (1989a) and others argue strongly that
underdetermination of this kind is the main problem with input.
One way in which this problem might be overcome is if the input provides
the child with negative evidence. Logically, there are two kinds of evidence,
positive and negative. Positive evidence comes from exposure to the speech of
other speakers, but as we have seen, this is not adequate because it un¬
derdetermines the final grammar. It follows that if children are to learn on the
basis of input alone they must receive negative evidence, i.e. be given feedback
that shows them what is ungrammatical in their sentences. However, as we
saw in Chapter 6, children do not typically receive direct negative feedback
on the grammaticality of their utterances. If there is no negative feedback,
how do children learn that sentences like:
 Linguistic universals and second language acquisition 435

*What did Randy wonder whether his brother would win?

are not possible sentences? The answer is, that with insufficient positive evid¬
ence and no negative evidence, they must rely on innate knowledge.
Another possible answer is that children act in accordance with indirect
negative evidence. That is, they avoid certain kinds of errors because they
never hear anyone produce them. The problem with this argument is that
there is ample evidence to show that children do produce errors that they
could never have heard in the input. Thus, as White points out ‘we would
need a theory which would explain why children notice the non-occurrence
of some sentence types but not others’ (1989a: 15). Such a theory would lead,
in fact, to claims that children have innate knowledge that guides them in
what to avoid. Thus, arguments based on indirect negative evidence cannot
replace those based on innateness.
As White puts it, ‘plausible theories of language acquisition must assume
realistic input’ (1990:124). From the preceding arguments two points follow:
(1) input alone cannot explain LI acquisition, and (2) therefore the child must
be equipped with knowledge that enables the deficiencies of the input to be
overcome. We should note, however, that not all the arguments relating to in¬
sufficiency of input, in particular those concerning negative evidence, neces¬
sarily apply to L2 acquisition, a point we will take up later (see page 456).

The language faculty

The child overcomes input deficiencies with the help of the language faculty.
This consists, in part, of UG, which assists the child in various ways. First, it
ensures that relatively little evidence is needed for the child to determine that a
given principle is operative in the target language or to decide which setting of
a parameter is the right one. Second, it prevents children from constructing
wild grammars (Goodluck 1986). That is, at no point does the child construct
a rule that contravenes UG. In this way, the child does not have to unlearn cer¬
tain types of errors, for which negative evidence would be necessary. Of
course, children do produce errors in features like 3rd person -s (for example,
‘Mommy like cake’), but these are ‘benign’ in the sense that they can be un¬
learned on the basis of positive evidence. Such structures involve language-
specific properties and, therefore, are not governed by UG. ‘Impossible
errors’ (errors that require negative evidence) do not need to occur precisely
because they are prohibited by UG. In essence, then, a language is learnable
because the child needs to entertain only a small subset of the hypotheses that
are consistent with the input data. Without the constraints imposed by UG,
LI acquisition would be at best extremely slow and, in some respects,
impossible.
There are two possibilities regarding the child’s access to the contents of
UG. One is that the entire contents are available from the start, as suggested
by White (1981). The other is that they are subject to maturation, with differ-
436 Internal factors

ent principles and parameters becoming accessible at different stages of ac¬

quisition, as proposed by Felix (1984).8 The problem with the first alternative
is that clearly children do not instantly arrive at the grammar of the target lan¬
guage. However, White argues that it can be assumed that there is an inter¬
action between UG and the language parser (the mechanism responsible for
processing input data), such that a certain principle is only triggered when the
child is capable of perceiving the relevant input data. White’s position is the
one preferred by many generative theorists because it suggests that the devel¬
opmental nature of LI acquisition is the product of non-linguistic rather than
linguistic factors, a point we will pick up later when we consider the domain
of a theory of L2 acquisition.
Some theorists have suggested that the language faculty contains learning
procedures or principles that are not properly part of UG but which operate
in tandem with it to enable the child to eliminate or avoid errors. An example
of such a principle is the Subset Principle (Berwick 1985). If learners construct
a conservative grammar, Y, which generates a subset of the sentences gener¬
ated by some other grammar, X (as shown in Figure 10.3), this grammar can
subsequently be expanded on the basis of positive evidence. However, if
learners began with a superset grammar, X, they would require negative evid¬
ence to narrow its scope in order to construct Y.9 For example, if children
initially construct a grammar that only allows NP + PP for dative verbs and
thus only produce sentences like:

Randy explained the problem to Mary,

they can avoid making impossible errors, as in:

* Randy explained Mary the problem.

As a result they will require only positive evidence to discover that some verbs
(like ‘give’) also permit NP + NP. In this way, they can overcome the limita¬
tions of the input—by avoiding, for example, the need for negative evidence.

Figure 10.3: The Subset Condition for two grammars (White 1989a: 143)

Chomsky views the language faculty as a mental organ, analogous to the

liver or the heart. He sees it as ultimately related to physical aspects of the
brain. Thus, ‘in certain fundamental respects we do not really learn language;
rather grammar grows in the mind’ (1980a: 134). The language faculty is a
child’s biological inheritance.
 Linguistic universals and second language acquisition 437

The logical problem of language acquisition

To conclude this section, we will state the underlying problem which Chom¬
sky’s theory of language has been designed to address. The logical problem of
language acquisition concerns how all children come to acquire with ease and
complete success a rich and complex body of linguistic knowledge despite
both their lack of cognitive sophistication and the poverty of the stimulus.
The answer lies in the language faculty—the principles and parameters of
UG, and learning procedures such as the Subset Principle.

The domain of a theory of second language acquisition

On the grounds that the same arguments used in support of UG and learning
principles are also applicable to L2 acquisition, it is possible to specify in
fairly narrow terms what it is that a theory of L2 acquisition needs to explain.

Grammatical competence

Essentially, the theory will need to account for the L2 learner’s grammatical
competence. That is, the theory should seek to account for ‘knowledge’ rather
than ‘behaviour’ or performance). It should also focus on ‘knowledge’ rather
than ‘ability’, and therefore need pay no attention to the variability inherent
in learner language, as this is a reflection of learners’ capacity to use their
knowledge in communication. Finally, it should limit itself to explanations of
the rules needed to account for the learner’s formal knowledge of L2 gram¬
mar, and need not consider how form-function relationships are estab¬
lished.10 ‘Knowledge’ here refers to knowledge that is implicit (unconscious
and intuitive) rather than explicit (conscious and metalingual).
Generative theorists, including Chomsky, have acknowledged that it is
possible to talk of pragmatic competence in the sense that speakers also in¬
ternalize a set of rules that govern how language is used to construct discourse
and to perform speech acts in socially appropriate ways. However, they typic¬
ally exclude this too, not because an account of pragmatic competence is not
needed, but because it is considered to be separate from grammatical compet¬
ence and because it is still poorly understood. As Gregg (1989) puts it:

... in comparison with an attempt to construct a theory of acquisition in

the domain of grammar, any attempt to construct a theory of acquisition in
the domain of pragmatics or communication is going to be handicapped by
the lack of a well-articulated formal characterization of the domain (1989:
24).

Pragmatic competence, therefore, is excluded on practical grounds.

The theory of L2 acquisition which generative theorists wish to see de¬
veloped is one that focuses on knowledge of grammar, defined as syntax,
phonology, and certain aspects of semantics. Such a theory is tenable if it is
438 Internal factors

accepted that (1) grammar constitutes an autonomous area of language, and

(2) the theory is modular in nature. These are seen as the essential character¬
istics of UG-based theories of L2 acquisition.

The autonomy of grammar

The claim that grammar constitutes an autonomous body of knowledge fol¬

lows from the hypothesis that speakers possess a language faculty that is inde¬
pendent of other cognitive systems such as those responsible for perception,
problem-solving or memorization. The evidence cited in support of this claim
comes from the highly abstract, specific, and precise nature of the principles
and parameters of UG. As Cook (1988) points out, ‘to defeat the argument in¬
volves explaining how each and all of these principles could have been ac¬
quired from experience or from other faculties’ (1988: 71). The generative
argument is that no such explanation has been or can be given.
It has also been claimed that the language faculty exists independently of
the mechanisms involved in using grammatical knowledge in performance.
This has led Cook (1985) to distinguish ‘acquisition’ and ‘development’. ‘De¬
velopment’ refers to real-time learning of a language and is influenced by non-
linguistic factors such as channel capacity. ‘Acquisition’ is language learning
unaffected by maturation, and is therefore dependent entirely on the learner’s
language faculty. Studies based on the analysis of performance data (for ex¬
ample, naturally occurring speech) only provide evidence of ‘development’.
In Chomsky’s words, they are ‘concerned with matters that may not properly
belong to the language faculty’ (1980a: 53) and provide evidence only of the
‘fluttering notions of a fledgling before the organs of flight mature’ (1981a:
35). It follows from such a view that to study ‘acquisition’ it is necessary to
adopt a logical rather than an empirical approach, through a consideration of
the relationship between the child’s ‘initial’ and ‘final state’.
According to Cook, therefore, ‘acquisition’ can only be examined in terms
of ‘an idealized “instantaneous” model in which time and experience play
minimal roles’ (1988: 81). For Cook, notions like ‘order’ and ‘sequence’,
which, as we have seen, are the focus of much L2 research, have no clear place
in such a model, as these are considered to reflect the learner’s developing
channel capacity rather than ‘acquisition’ itself. The view held by White
(1981), that the entirety of UG is available to the child from the start and that
‘development’ is a reflex of improving perceptual abilities, also appears at
first sight to exclude considerations of order and sequence. However, not all
researchers working within a UG framework reject developmental data. As
we noted earlier, Felix (1984) claims that ‘the principles of Universal Gram¬
mar are themselves subject to an innately specified developmental pro¬
cess’—they emerge rather like teeth, in a pre-determined sequence. Also, a
non-maturational view of UG, such as that preferred by White, still allows for
a consideration of ‘learning difficulty’. This is definable in terms of how a
 Linguistic universals and second language acquisition 439

static language faculty responds to the available input, with unmarked fea¬
tures requiring less elaborate triggering experience than marked ones (see the
section on markedness on pages 430-3). As Hyams (1991) points out, there
are in fact two problems that need to be addressed: the ‘logical problem’,
which ‘treats acquisition as an instantaneous process’, and the ‘develop¬
mental problem’, which recognizes that acquisition is non-instantaneous. As
we will see, UG-oriented SLA research has addressed both problems.

Modularity

In claiming that grammar is autonomous, generative theorists recognize that

other domains, such as perception and memory, are involved in language ac¬
quisition, and that these interact with the language module. In other words, a
theory of how linguistic competence is acquired constitutes only one module
in an overall theory of L2 acquisition. Furthermore, even within the language
module itself, it may be necessary to identify various sub-domains, including
UG, language learning procedures or principles, the domain responsible for
the language specific learning needed to acquire peripheral grammatical fea¬
tures and vocabulary, and a language parser that analyses input (see Figure
10.4, from White 1989a). A complete theory, therefore, will have to explain
how all the modules and sub-modules shown in Figure 10.4 interact.
We are a long way from being able to do this. The generative solution is to
focus, narrowly and perhaps sensibly, on a specific domain. In so doing, gen¬
erative theorists are at pains to acknowledge that the explanation they pro¬
vide is only a partial one. The claim is only that ‘UG plays a central and vital
part in L2 learning, but there are many other parts’ (Cook 1988: 189).

LANGUAGE MODULE CENTRAL PROCESSORS

UG Grammar Memory

Beliefs

Pragmatics
Language
Language
learning Real-world knowledge
Parser
principles
Problem-solving
abilities
etc.
PERCEPTUAL MODULES
Vision, hearing, etc.

Figure 10.4: Cognitive modularity (from White 1989a: 178)

440 Internal factors

Empirical studies of second language acquisition based on Universal

Grammar
It was not until after the publication of Chomsky’s Lectures on Government
and Binding in 1981 that linguistic theory began to attract the serious atten¬
tion of L2 researchers. Since then studies have proliferated. No attempt will
be made to survey them in this chapter. Instead, representative studies that
have examined the following key issues will be examined:

1 Do L2 learners have continued access to UG?

2 Do learners experience less difficulty with unmarked than with marked
features of the L2?
3 Do L2 learners make use of the same learning principles as LI learners?
Although a number of studies (for example, Hilles 1986; 1991) have studied
learner language, much of UG-based SLA research has been directed at identi¬
fying learners’ intuitions, and to achieve this has developed its own research
methodology. It is appropriate to begin this section with a few comments
about this.

The methodology of UG-based studies

Second language researchers basing their research on UG need to address sev¬
eral methodological issues (see White 1989a). One of these is how to ensure
that the subjects have the requisite level of L2 proficiency to demonstrate
whether or not a particular principle is operating in their interlanguage gram¬
mar. Many of the principles identified by UG grammarians involve complex
sentences, and can therefore only be expected to manifest themselves in the
later stages of development. White argues that it is essential to ensure that
subjects are able to handle the necessary structures otherwise ‘learners might
violate a universal not because of the non-availability of UG, but because the
structure in question is beyond their current capacity’ (1989a: 61).
Another problem concerns the need to rule out the effects of the LI. If sub¬
jects act in accordance with UG, this might be because they have direct access
to its contents or because they have indirect access through their LI. Thus, it
is necessary to investigate learners with Lis that do not manifest the specific
principle in question. Luckily this is possible, because not all UG principles
operate in all languages. Subjacency, for example, does not operate in wh-
-questions in languages like Chinese, Japanese, and Korean (see page 443).
Thus, if learners with these Lis follow this principle in an L2 like English,
which does manifest it, this provides clear evidence that they are acting in ac¬
cordance with UG constraints—providing it can be shown that they have not
obtained knowledge of this principle through formal instruction.
In the case of studies of parameter setting, where the aim is to establish
which setting out of those possible is reflected in the learner’s interlanguage,
bi-directional studies of the kind found profitable in language transfer studies
 Linguistic universals and second language acquisition 441

(see Chapter 8) are needed. For example, the pro-drop parameter requires a
study that investigates how English learners of L2 Spanish and Spanish learn¬
ers of L2 English handle the relevant structural properties. Such studies are
made even stronger if there are control groups where the LI and L2 share the
same parameter setting (for example, French and English). Only in this way is
it possible to reach conclusions regarding the inter-related effects of transfer
and markedness on acquisition.
Perhaps the most controversial methodological issue in UG-based studies
concerns what kind of L2 data to collect. This is, of course, a problem for all
L2 acquisition researchers but it is especially problematic in the case of E1G-
based research because of the need to obtain information about learners’
competence rather than their ability to perform specific structures. There is
also the difficulty of obtaining samples of language use that contain the kinds
of complex structures needed to investigate most principles and parameters.
For both these reasons, elicited data have been preferred.
White (1989a) mentions a number of methods for collecting data. These in¬
clude the use of act-out tasks, picture identification tasks, sentence-joining
tasks, and card-sorting tasks. Flowever, the main method used to date is the
grammaticality judgement task. This comes in several forms (see Ellis 1991b
for an account of these), but always involves the learner in making some kind
of metalingual assessment regarding the grammaticality of a mixed set of sen¬
tences, some grammatical and some ungrammatical. As White points out, the
great advantage of this kind of task is that it forces subjects to consider sen¬
tences that are ‘impossible’ from a UG standpoint. Rejecting such sentences
indicates that FIG is alive, while accepting them shows that it is dead.
There are many problems with grammaticality judgements, however. Bird¬
song (1989) points out that they are not appropriate for learners with poor L2
literacy, and that differences in the metalinguistic skills of literate learners are
also likely to affect responses. Furthermore, when learners reject sentences it
is not always clear whether this is because of their grammatical properties or
because of the difficulties that they experience in trying to parse them, a point
taken up later. Birdsong also points out that often learners lack confidence
and, therefore, are reluctant to commit themselves to a definite judgement, a
point borne out in a study reported in Ellis (1991c). I tested different groups
of Chinese and Japanese learners of English on two occasions, one week
apart, using sentences with dative verbs (for example, ‘show’ and ‘explain’),
and found that the subjects frequently changed their judgements (up to 46 per
cent of the time in one case). Variability in learners’ judgements is therefore a
major problem because it casts doubt on the reliability of the grammaticality
judgement test.
Many of the problems that I and Birdsong raised may be overcome by bet¬
ter designed tests and by retesting to check reliability. Flowever, in one other
respect grammaticality judgements seem insuperably problematical. As Bia-
lystok’s (1979) study indicates, L2 learners make use of both implicit and
442 Internal factors

explicit knowledge in reaching judgements. Indeed, the very nature of the

grammaticality judgement task encourages the use of explicit knowledge.
However, as was made clear in the previous section, UG is a theory of implicit
linguistic knowledge. It might be argued that learners are not likely to have
explicit knowledge relating to sentences that violate UG principles, but, as
several studies have shown (for example, Seliger 1979; Sorace 1988), learn¬
ers’ explicit knowledge is often anomalous, with the result that their judge¬
ment of any sentence, including ‘impossible’ ones, may be uncertain and
inconsistent. Unless some way can be found to ensure that learners do not use
their explicit knowledge, it is not clear how the data obtained from such a
task can be used to make claims about the role of UG in L2 acquisition.11
UG-based SLA research is typically experimental in nature: it involves con¬
trol and experimental groups and elicited rather than naturally occurring lan¬
guage behaviour. This is because UG affords very precise hypotheses about
the nature of L2 acquisition which lend themselves to an experimental treat¬
ment. Also, unlike much of the earlier research, where the primary goal was
the description of learner language, UG-based research is explanatory in
nature.

The availability ofUG in second language acquisition; some

empirical research

There has been extensive theoretical debate regarding the availability of UG

in L2 acquisition. However, we will delay consideration of this until a later
section (see page 453). Here we will examine what light empirical research
has shed on this issue, focusing our attention on studies that have investigated
the Subjacency Principle and the pro-drop parameter. (See page 430 in this
chapter for a definition of subjacency, and White 1989a for an extensive
review of the research.)

Subjacency

In an early study, Ritchie (1978b) investigated the ‘right roof constraint’, a

principle that in more recent grammatical models falls under Subjacency. The
right roof constraint limits the extent to which ‘rightward movement’ is pos¬
sible, allowing sentences like:

A man in a car has just passed by.

A man has just passed by in a car.
That a man has just passed by in a car was not noticed.

but preventing sentences like:

’■'That a man has just passed by was not noticed in a car.

Ritchie investigated twenty adult Japanese graduate students at an American

university and six native speakers by asking them to judge pairs of sentences,
 Linguistic uniuersals and second language acquisition 443

one of which conformed to the right roof constraint and the other of which
did not. Japanese is a language with no rightward-movement rules. The ma¬
jority of the subjects generally judged the sentences that violated the con¬
straint to be ‘less grammatical’ than the sentences that did not (i.e. they
performed significantly better than chance). However, there were some sub¬
jects whose judgements did violate the constraint. Ritchie’s tentative conclu¬
sion was that the study provided ‘preliminary support to the assumption ...
that linguistic universal are intact in the adult’ (1978b: 43).
Further evidence for adult learners’ continued access to UG comes from a
study by Bley-Vroman, Felix, and Ioup (1988). The subjects of this study
were 92 advanced Korean learners of L2 English, most of whom had been liv¬
ing in the United States for several years, and as a result of which had had
plenty of opportunity to learn naturalistically. Korean is not constrained by
subjacency in ^-questions. The grammaticality judgement test required the
subjects to indicate whether sentences were grammatical or ungrammatical,
or whether they were not sure. The sentences, all of which involved wb-
movement, were designed to test access to the Subjacency Principle. There
were also a number of sentences which did not involve the principle and
which were used as controls. The results showed a response bias (the learners
manifested a tendency to reject sentences irrespective of whether they were
grammatical or ungrammatical) and a reluctance to make use of the ‘not sure’
option. However, they also showed that the learners were not guessing at ran¬
dom; the typical response pattern resembled that of the native speakers. Bley-
Vroman et al. conclude: ‘Given these results, it is extremely difficult to main¬
tain the hypothesis that Universal Grammar is inaccessible to adult learners’
(1988:26).
Further evidence for adult learners’ continuing access to UG comes from a
study by White, Travis, and Maclachlan (in press). This study looked at
Malagasy learners of English, Malagasy being a language that permits subject
extraction in u^-interrogatives where English does not. Thus, if the learners
rejected sentences manifesting subject-extraction, this could not be explained
by LI transfer. Results from a grammaticality judgement test and a written
elicited production task showed that nearly all the high-intermediate learners
and half of the low-intermediate learners rejected Subjacency violations.
White et al. argue that those subjects who accepted sentences that violated
this UG principle did so only because they had not reached a stage of syn¬
tactical development for the Subjacency Principle to become active (i.e. they
had not acquired z^-movement yet).
Whereas Ritchie’s, Bley-Vroman et al.’s, and White et al.’s studies all sup¬
port the ‘UG is alive’ position, Schachter’s (1989) study indicates the oppos¬
ite. Schachter chose as subjects learners whose LI did not reflect subjacency in
^-questions (Korean), learners whose LI provided only weak evidence of
subjacency-based rules in ^-questions (Chinese), and learners whose LI
clearly manifests subjacency but not in ^-movement (Indonesian). In
444 Internal factors

addition, there was a group of native speakers of English to act as controls.

Schachter administered a grammaticality judgement test designed to assess
whether the subjects had developed sufficient knowledge of the grammatical
structures involved, in addition to the tests intended to show whether they
violated the Subjacency Principle. The native speakers passed both the syn¬
tax and the subjacency tests. However, many of the learners passed only the
syntax test—they exhibited knowledge of a syntactic construction without
corresponding knowledge of the subjacency constraint. The behaviour of the
learners was significantly different from that of the native speakers. There
were also some differences among the different groups of learners but these
were not significant. Thus, even if subjacency was instantiated in the learners’
LI, it did not guarantee access to it in the L2. Schachter considers that the res¬
ults constitute ‘a serious challenge’ to the claim that UG is available to adult
L2 learners.
The picture is further complicated by uncertainty regarding the nature of
learners’ judgements when asked to judge the grammaticality of sentences
like:

What did Bill think that the teacher had said?

*What did Sam believe the claim that Carol had bought?
(from Bley-Vroman et al. 1988: 8).

Bley-Vroman et al. note that the results they obtained may have reflected ease
of parsing rather than UG effects. This hypothesis has been given support by a
study by Schachter and Yip (1990). This produced clear evidence of pro¬
cessing effects in both native speakers’ and learners’ judgements of sentences
involving wh-movement. Both sets of learners found it more difficult to make
accurate judgements in sentences with three clauses than in sentences with a
single clause. Schachter and Yip suggest that this is because subjects have to
keep the wh-word in short-term memory until they discover the clause in
which it fits. The subjects also performed much less accurately with sentences
where the wb-extraction involved grammatical subjects than when it in¬
volved objects, irrespective of the number of clauses involved. In other words,
both native speakers and learners treated grammatical extraction from object
and subject positions differently. As this cannot be explained by reference to
UG, it provides further evidence of processing factors at work. Schachter and
Yip conclude that ‘a number of earlier judgmental studies may need to be
reexamined in terms of the possible effect of processing constraints’, and that
researchers need to be circumspect in accepting ‘pure grammatical
explanations’.12
A more recent study by Uziel (1993), however, suggests that learners’
greater difficulty with subject extraction is explicable in terms of UG. Draw¬
ing on the grammatical arguments which Chomsky (1986b) presented in Bar¬
riers (a later framework than the Government/Binding model used in the
earlier studies), Uziel hypothesized that subject-extraction would be rejected
 Linguistic universal and second language acquisition 445

at a higher rate than object-extraction because it violates two principles (Sub-

jacency and another), whereas object-extraction violates only Subjacency. He
presents results from a small-scale study involving grammaticality judge¬
ments by 11 adult Italian and 10 adult Hebrew ESL learners that support this
hypothesis. Uziel’s study suggests that UG principles can help to explain why
learners find some syntactic constructions more acceptable than others.
Given the problems that learners appear to experience with judging arcane
sentences of the kind often found in the tests used in these studies, it is perhaps
safer not to try to reach any conclusions regarding the availability or non¬
availability of UG.13 Whether adult learners have access to the Subjacency
Principle clearly requires further study directed at ‘relative acceptability’
rather than ‘absolute obedience’ to UG, as suggested by Uziel, and, ideally,
involving data that are not drawn exclusively from grammaticality judge¬
ment tasks.

The pro-drop parameter

Another way of testing for the availability of UG is by investigating whether

adult L2 learners are capable of resetting a parameter from the value found in
their LI to the value of the L2. If resetting does take place, this can be taken as
evidence for the continued existence of UG. A second issue of considerable in¬
terest is whether learners reset all the features associated with a parameter
(i.e. whether clustering effects are evident). We will focus here on studies
based on the pro-drop parameter.
Two studies by White (1985; 1986) indicate that initially L2 learners opt
for the LI setting of the pro-drop parameter, but that as their proficiency in¬
creases they switch to the L2 setting. White also investigated whether pro¬
drop features cluster in interlanguage grammars. The subjects of the 1985
study were Spanish and Lrench learners of L2 English, while the 1986 study
also included two Italian learners. Spanish and Italian are pro-drop languages
while Lrench and English are non-pro-drop. The learners were asked to judge
sentences, including some with missing pronoun subjects (both expletives, ‘it
and ‘there’, and referential pronouns, such as ‘he’ and ‘she’), some with un¬
grammatical subject-verb inversion, and some with a ‘that’ trace. White
found clear evidence of differences in the performance of the Spanish and Ital¬
ian learners compared to the Lrench. They were more inclined to accept sub¬
jectless sentences. Also, of those subjects who demonstrated the necessary
syntactical development, the Spanish learners were more likely to accept
‘that’-trace than the Lrench. However, there was no difference in their judge¬
ments of sentences with ungrammatical subject-verb inversion. The accuracy
level of the learners’ judgements improved in accordance with their overall
proficiency. These results suggest the following conclusions: (1) L2 learners
do not interact directly with L2 data but, instead, initially transfer the LI set¬
ting of a parameter, (2) given time, learners succeed in resetting a parameter
446 Internal factors

to the new L2 value, and (3) target-language features may cluster in

interlanguage grammars, although not entirely as predicted by linguistic
theory.14
Whereas White’s research was based on grammaticality judgements and, in
one of the studies, elicited question production, the studies to be considered
next have all made use of more naturally occurring learner language. Given
the doubts that exist regarding the reliability and validity of grammaticality
judgement tasks, they provide an important alternative source of evidence.1’
Hilles (1986; 1991) made use of the longitudinal data collected by Cancino
et al. (1978) in their study of the naturalistic L2 acquisition of English nega¬
tives and auxiliaries, by Spanish speakers (see Chapter 3). In the 1986 study,
Hilles looked at just one learner, Jorge, a 12-year-old Colombian. Hilles hy¬
pothesized that Jorge would begin with the LI setting (pro-drop), and sub¬
sequently switch to the L2 setting (null pro-drop). She further hypothesized
that this switch would co-occur with the emergence of auxiliary, and that it
would be triggered by the acquisition of the expletives ‘it’ and ‘there’. Her
analysis supported these hypotheses, although Hilles was careful to point out
that more evidence was needed before firm claims could be made.
The 1991 study focused on two of the features associated with the pro¬
drop parameter—the use of pronominal subjects and verb inflection—and
sought to establish to what extent the two were correlated over time.16 In this
study, all six of Cancino et al’s subjects (two children, two adolescents and
two adults) were included, thus allowing Hilles to investigate to what extent
age was a factor in the availability of UG. Three of the learners—the two
children and one adolescent (Jorge)—manifested a strong correlation
between the emergence of pronominal subjects and verb inflection,
suggesting that their acquisition was guided by UG. Hilles argues that the
developmental sequence these learners followed mirrored that found in LI
acquisition. LI learners begin with null subjects and uniformly uninflected
verbs, and subsequently switch to pronominal subjects once they realize that
English is not uniform with regard to verb inflection (see Jaeggli and Hyams
1988). In contrast, the other adolescent and the two adults displayed no such
correlation, indicating that they lacked access to UG.
A problem of Hilles’ studies is that because there are no controls (i.e. sub¬
jects with non-pro-drop Lis like Lrench), it is impossible to tell whether her
Spanish subjects’ early preference for the pro-drop setting is a true reflection
of LI acquisition or the result of LI transfer. Lakshmanan’s (1991) study,
however, allows us to consider this question, at least in the case of children.
Lakshmanan examined longitudinal data for three learners: Marta (one of
the children in Cancino et al.’s study), Muriel (a Lrench-speaking child stud¬
ied by Gerbault 1978) and Uguisu (the Japanese child studied by Hakuta
1974). Whereas Spanish and Japanese can be considered pro-drop languages,
Lrench, like English, is non-pro-drop. The results support neither a ‘transfer’
nor a ‘developmental’ explanation, nor do they provide unequivocal evidence
 Linguistic universals and second language acquisition 447

in favour of the clustering effect Hilles reported. Marta began by using null
subjects, but these rapidly gave way to pronominal subjects. However, Marta
never mastered English verb inflections in the period covered by the study.
Muriel’s use of null subjects was restricted to contexts involving ‘it is’ and,
again, there appeared to be no relationship between her use of pronominal
subjects and verb inflections. Uguisu did not employ any null subjects, but,
contrary to the claims of the pro-drop parameter, treated English as morpho¬
logically uniform in the early stages. As Lakshmanan points out, these results
are not readily explicable in terms of UG.17
Phinney’s (1987) study also compared learners with different Lis, once
again focusing on the presence and absence of subject pronouns and the verb
inflectional system. In this case, however, the study was bidirectional in
nature, comparing English-speaking learners of L2 Spanish and Spanish¬
speaking learners of L2 English. The data for this study came from written
compositions. All the learners were fairly accurate in subject-verb agreement,
but they differed in their use of subject pronouns. The L2 English learners
omitted few referential pronouns but many expletive pronouns (‘it’ and
‘there’), suggesting that they might have transferred the LI value of the para¬
meter. The L2 Spanish learners, however, provided no evidence of transfer,
correctly omitting both referential and expletive pronouns. As White (1989a)
points out, it is difficult to reach clear conclusions on the basis of this study as
the two groups of subjects may not have been equivalent.
Overall, these studies provide no real support for a parameter-setting
model of L2 acquisition. One general finding is that learners with pro-drop
Lis tend to omit subject pronouns in the L2 to begin with and then later learn
to include them (although Lakshmanan’s study provides counter-evidence).
But it is not clear that this requires a UG explanation; it is not necessary to in¬
voke ‘parameter-setting’ to explain why L2 learners with pro-drop Lis pro¬
duce sentences with no subject pronouns in the L2. Nor, on the basis of these
studies, is it possible to conclude that UG works in child L2 acquisition but
not in adult. Nor is there clear evidence of any clustering effect. White’s and
Hilles’ studies suggest that there is some degree of clustering in the acquisition
of the features hypothesized to be related to the prodrop parameter, although
this does not always conform to the predictions of the linguistic models on
which these studies were based. Other studies (Lakshmanan’s and Phinney’s),
however, provide no evidence of clustering.

Markedness in UG and second language acquisition

We turn now to considering some of the empirical research that has examined
the relationship between markedness and L2 acquisition. This research was
based on the UG-based definitions of markedness considered earlier.
It will be recalled that in UG theory, ‘core’ rules are unmarked and ‘peri¬
pheral’ rules marked (see page 432). Mazurkewich (1984; 1985) investigated
448 Internal factors

dative alternation, arguing that the NP + PP pattern is unmarked and the NP +

NP pattern marked. One reason she gives is that the former is more product¬
ive than the latter (i.e. almost all dative verbs take NP + PP complements
while only some take NP + NP). Another reason is that case assignment is
transparent in the NP + PP pattern but is problematic in the NP + NP pattern.
Mazurkewich considers the NP + PP complement to be a core rule and NP +
NP a peripheral rule. She hypothesized that sentences like:

John baked a cake for Mary. (-NP + PP)

would be easier to acquire than sentences like:

John baked Mary a cake. (-NP + NP)

The subjects in Mazurkewich’s (1984) study were 45 French-speaking high

school and college students and 38 Inuktitut-speaking high school students.
They were divided into beginner, intermediate, and advanced levels on the
basis of scores on a cloze test. There was also a group of native-speaker con¬
trols. Judgements about sentences containing dative verbs with both the NP +
PP and the NP + NP patterns were elicited from all the learners and the con¬
trols. Mazurkewich found that both the French and Inuktitut speakers judged
the sentences with the unmarked NP + PP pattern more accurately than the
sentences with marked NP + NP. She also found that the level of accuracy in
the marked pattern increased with proficiency. Taken together, these results
suggest that the learners acquired the unmarked structure first.
There have been several criticisms of Mazurkewich’s research, however.
Kellerman (1985b) points to a number of design flaws. White (1989a) notes
that the French learners may have been influenced by their LI (which permits
only the NP + PP pattern) rather than by markedness, and she also notes that
the results obtained for the Inuktitut speakers were very similar to those ob¬
tained for the native speakers. She argues, therefore, that the study does not
provide any evidence for an acquisition sequence determined by markedness.
Hawkins (1987) goes further, challenging Mazurkewich’s notion of marked¬
ness. He points out that dative alternation is generally seen as a lexical prop¬
erty of verbs and that lexical properties belong to the periphery rather than
the core. He argues that, according to current linguistic theory, dative al¬
ternation cannot be used as an example of the core/periphery distinction.
Hawkins’ own study of dative alternation investigated ten adult French
speakers spending a year at a British university. They were given a grammat-
icality judgement task similar to Mazurkewich’s and a sentence construction
task that required them to add a preposition (‘to’ or ‘for’) to any sentences
they thought required them. Through a highly detailed analysis of the results,
Hawkins provides evidence of a series of stages in the acquisition of dative al¬
ternation, reflecting the progressive introduction of syntactic features into the
interlanguage grammar. To begin with, learners distinguish dative verbs ac¬
cording to whether they take a pronominal or a lexical object (for example,
 Linguistic universals and second language acquisition 449

.. give Mary it’ v. . give a present to Mary’). Later they introduce a dis¬
tinction between ‘to’ and ‘for’ verbs (for example, ‘give’ and ‘cook’), while
later still they distinguish native and non-native verb forms (for example,
‘give’ and ‘donate’). These stages were evident in individual dative verbs
rather than across the board, suggesting that there was ‘a progressive spread
through the verbs of the learner’s lexicon’ (1987: 24). On the basis of these
findings, Hawkins argues that it is misleading to isolate a single aspect of dat¬
ive alternation, as Mazurkewich does, and raise it to the status of a UG-
determined feature of L2 acquisition. His own preference is for a theory of
markedness based on cognitive notions of ‘learning complexity’.
There have been several other studies of the role of UG-derived markedness
in L2 acquisition, but in just about every case they face similar criticisms to
those levelled against Mazurkewich’s research. The main problem rests in the
twists and turns of linguistic theory. Noting this, Van Buren and Sharwood
Smith (1985) ask ‘when linguistics coughs, should second language acquisi¬
tion catch pneumonia?’ (1985: 21). They go on to answer this question by
claiming that the theoretical foundations (i.e. Chomsky’s Government/
Binding Theory) are sufficiently robust to warrant application in SLA re¬
search. This is not a view shared by all researchers however, (for example,
Hawkins). Lack of consensus about the details of the theory is often evident,
as, for instance, in the debate centring on Flynn’s research on the head-final/
head-initial parameter (see Flynn 1987, Bley-Vroman and Chaudron 1990,
and Flynn and Lust 1990). It is clearly premature to reach any conclusions as
to whether markedness, as defined by the theory of UG, is a relevant factor in
L2 acquisition.

Learning principles in second language acquisition

Earlier we noted that some theorists have posited the need for learning prin¬
ciples, such as the Subset Principle, to ensure that learners do not build gram¬
mars with rules that cannot be disconfirmed solely on the basis of positive
evidence (see page 436).
White (1989b) investigated the L2 acquisition of adverb placement by ad¬
ult French-speaking learners of English.18 Whereas French allows adverbs to
be positioned between the verb and the direct object:

Marie a mange rapidement le diner.

English does not:

"‘Mary ate rapidly her dinner.

White argues that the Subset Principle should lead the learners to opt for a
grammar that excluded adverb placement between verb and direct object (the
conservative option). LI transfer, on the other hand, would result in an L2
grammar that allowed this placement.
450 Internal factors

Three different data collection tasks were used: a paced grammaticality

judgement test, an unpaced multiple choice grammaticality judgement test,
and a preference task (where the subjects were asked to examine pairs of sen¬
tences and indicate which seemed ‘better’). These tests were administered to
the 43 learners and to a group of native-speaker controls. The results showed
that the learners were much more likely to accept the ungrammatical sen¬
tences (i.e. those where the adverb was placed between verb and direct object)
than the native speakers in the grammaticality judgement tests, and were
much more likely to rate the grammatical and ungrammatical sentences as the
same in the preference test. White concludes that these results support a trans¬
fer rather than a subset hypothesis.
This study and others (for example, Zobl 1988; Thomas 1989) indicate
that the Subset Principle is non-operative in L2 acquisition. L2 learners ap¬
pear to construct a superset grammar when such a grammar is suggested by
their LI. This raises the question as to whether learners are subsequently able
to readjust to a subset grammar. Evidence that they are able to reconstruct
their grammars in this way would indicate that UG is still available to them
(i.e. they have access to new, non-Ll parameter settings). Conversely, evid¬
ence that they cannot do so would suggest that they are stuck with the LI set¬
tings of parameters because UG is not available. In essence, this takes us back
to the same question we addressed when we considered parameter setting.
Studies by Thomas (1989) and Hirakawa (1989) suggest that even though
learners construct superset grammars, contrary to the Subset Principle, some,
at least, ultimately arrive at the target-language setting.
How, then, do L2 learners move from a superset to a subset grammar?
Logically, this is not possible on the basis of positive evidence. Indeed, the
Subset Principle was set up to account for the impossibility of eliminating
problematic errors by means of positive evidence. The French learner of L2
English can never be certain that sentences like:

" Mary ate rapidly her dinner.

are not permitted. One possible solution to this conundrum, proposed by

White (1989a), is that L2 learners make use of grammatical explanations and
negative evidence (in the form of corrections of their errors) to help them es¬
tablish what is grammatical and what is ungrammatical. In other words, L2
learners differ from LI learners because they do not have access to learning
principles like the Subset Principle, but they are able to overcome the diffi¬
culties that this gets them into by utilizing negative evidence not available to
the LI learner. Another possibility, of course, is that the Subset Principle is in¬
correct and the linguistic theory is wrong.
To sum up, the available research strongly suggests that L2 learners do not
have access to learning principles like the Subset Principle. Consequently,
they build a superset L2 grammar, often influenced by their LI. Subsequently,
however, they restructure this grammar by restricting rule application, thus
 Linguistic universals and second language acquisition 451

creating a subset grammar. It has been suggested by White that this requires
negative evidence.
This position affords several testable hypotheses. One is that LI learners
will not make certain kinds of error (for example, placement of an adverb be¬
tween verb and direct object) whereas L2 learners will. This hypothesis has
received some support from studies which have investigated another UG
principle (the Governing Category Parameter), which concerns what governs
the relationship between pronouns and their antecedents (see, for example,
Finer 1991). In LI acquisition, children appear to conform to the narrowest
value of this parameter, while in L2 acquisition, adults start off with the
widest value if that is the setting in their LI. In other words, LI acquisition
conforms to the Subset Principle, whereas adult L2 acquisition does not.
A second hypothesis is that L2 learners with no access to negative evidence
(for example, naturalistic learners) will fail to eliminate superset errors. This
has not been investigated. A third is that L2 learners who receive formal
instruction will eliminate them. This has received support and will be
considered further in Chapter 14.

Summary

This section has examined some of the L2 research based on the theory of UG.
Much of this research has made use of grammaticality judgement tasks,
which are problematic in a number of ways, in particular because we do not
know whether they tap implicit or explicit knowledge. The research does not
provide a clear answer as to whether UG is alive or dead in the L2 learner: stu¬
dies of L2 learners’ access to the Subjacency Principle and the pro-drop para¬
meter have produced mixed and indeterminate results. Similarly, there is no
clear evidence to support the hypothesis that learners acquire unmarked
‘core’ features before marked ‘peripheral’ features, at least in the case of dat¬
ive alternation. Clearer results have been obtained where learning principles
are concerned. L2 learners do not follow the Subset Principle, although some
still do ultimately arrive at a correct target-language grammar, perhaps be¬
cause they have access to negative evidence. Overall, although the studies
considered in this section have afforded a number of insights into how learn¬
ers handle formal grammatical properties, they have not provided clear
answers to the key questions, partly because of methodological problems and
partly because the details of the linguistic theory on which they have been
based have been unstable.

Theoretical issues
We return now to address the main theoretical issues in UG accounts of L2 ac¬
quisition. Given the indeterminacy of the results of empirical research, it is
452 Internal factors

not surprising to find quite contradictory views expressed. As Eubank

(1991a) so aptly puts it, the current state of is one of ‘point counterpoint’.

The logical problem of second language acquisition

A number of theorists have argued that the logical problem of language ac¬
quisition applies just as much to L2 acquisition as it does to LI acquisition.
Cook (1988), for instance, claims that like LI learners, L2 learners possess
knowledge of the L2 that they could not have acquired from the input and
which must, therefore, have existed within their own minds. In other words,
‘the poverty of the stimulus argument applies equally to L2 learning’ (1988:
176). He also argues that imitation, grammatical explanation, correction and
approval, social interaction, and dependence on other faculties cannot ac¬
count for L2 acquisition any more than they can for LI acquisition. Lor
Cook, the essential difference between LI and L2 acquisition is that L2 learn¬
ers already know another language, which may serve as a alternative source
to UG.
There are, however, several other significant differences between first and
second language learners and it is these that theorists like Bley-Vroman
(1989) and White (1989a) point to in arguing that the logical problem of lan¬
guage acquisition is somewhat different in the case of L2 acquisition. Bley-
Vroman identifies a number of differences (see Table 3.10) and on the basis of
these argues:

The logical problem of foreign language acquisition becomes that of ex¬

plaining the quite high level of competence that is clearly possible in some
cases, while permitting the wide range of variation that is possible (1989:
49-50).

The question arises as to whether the solution to this logical problem, which
is very different to that of LI acquisition, necessitates an innate domain-
specific acquisition system. Bley-Vroman’s answer is a definite ‘no’, but, as we
will see, other theorists have been more circumspect. White (1989a: 45), for
instance, answers with a ‘not sure’, pointing out that the differences do not re¬
quire the abandonment of UG-mediated L2 learning, while also noting that
the arguments in favour of it are not clear-cut.
The differences discussed by Bley-Vroman are essentially quantitative in
nature. But what if L2 acquisition can be shown to be qualitatively different
to LI acquisition? Schachter (1988) notes that even proficient L2 learners fail
to acquire movement rules such as rules relating to topicalization and adverb
placement.(‘Topicalization’ refers to ‘the use of various grammatical devices
for placing the topic of a sentence in sentence initial position (e.g. “As for
love, it is no substitute for money”)’). Schachter argues that many learners
achieve communicative fluency without complete grammatical mastery of the
L2. Schachter queries whether any adult L2 learner is capable of achieving a
 Linguistic universals and second language acquisition 453

mental state comparable to that achieved by a native speaker of the target

language. In other words, she sees the L2 learner’s grammatical competence
as different in nature from that of the LI learner. If this view of L2 acquisition
is accepted, the logical problem becomes very different—it is that of ex¬
plaining why LI and L2 competences are qualitatively different.
In summary, there are different positions regarding the logical problem of
L2 acquisition. One is that it is essentially the same as for LI acquisition. An¬
other is that it is different because L2 learners achieve variable success. A
third is that it is different because L2 competence is qualitatively different
from LI competence. These positions lead to different views regarding the
role of UG in L2 acquisition.

Access to UG in second language acquisition

A number of different views relating to the availability of UG in L2 acquisi¬
tion can be distinguished: these are (1) the complete access view, (2) the no ac¬
cess view (sometimes referred to as ‘the Fundamental Difference Hypothesis’,
(3) the partial access view, and (4) the dual access view. We will outline each
position before attempting an evaluation of them.
The complete access view is evident in the Parameter-setting Model of
Flynn (1984; 1987). Flynn argues that ‘the essential faculty for language evid¬
enced in LI acquisition is also critically involved in L2 acquisition’ (1987:
29). However, Flynn also acknowledges a crucial role for the LI. In cases
where the LI and L2 parameter settings are the same, learning is facilitated
because ‘these L2 learners are able to consult the structural configuration es¬
tablished for the LI in the construction of the L2 grammar’ (op. cit.: 30).
Where the LI and L2 parameter settings are different, the learner has to as¬
sign new values and, although this is not problematic according to Flynn, it
does add to the learning burden. Flynn (1987) hypothesizes that where the LI
and L2 have identical settings, the pattern of acquisition of complex sentence
structures (of the kind that UG principles typically address) will correspond
to the later stages of LI acquisition. She also hypothesizes that where the LI
and L2 have different settings, the pattern of acquisition will correspond to
the early stages of LI acquisition, as the learners need to first discover the rel¬
evant structural configuration in the L2.
Flynn’s parameter-setting model rests on the assumption that adult L2
learners have access to the same language faculty as LI learners.19 As such it
rejects the claim advanced by other theorists that age is a significant factor in
L2 learning (see Chapter 13 for a detailed discussion of this issue). Flynn and
Manuel (1991) explicitly address the age issue and conclude that ‘it is imposs¬
ible to argue for a monolithic critical period in L2 learning’ (1991: 140). They
present three arguments in favour of this position. First, like LI learners, L2
learners possess grammatical knowledge that could not have been learnt
purely on the basis of input. Second, L2 learners possess knowledge that is
454 Internal factors

structure-dependent. Third, they exhibit the same infinite productivity of new

sentences as LI learners. In essence, Flynn and Manuel are asserting that the
logical problems of LI and L2 acquisition are the same.
It is possible, of course, that adult L2 learners may have continued access to
UG, while the grammars that they build sometimes manifest properties that
violate UG principles. White (forthcoming) refers to this as the ‘different
competence, same mechanisms’ position, which might be considered a
weaker version of the complete access position. White advances the argument
that one reason why adult L2 learners may sometimes construct a grammar
that is not in conformity with UG is because their LI leads them to misanalyse
the input—they do not so much transfer an LI property as misinterpret the
input in terms of expectancies based on their LI.
A number of theorists support a no-access view (for example, Clahsen and
Muysken 1986; Meisel 1991). This position rests on two related claims. The
first is that adult L2 acquisition is very different from LI acquisition. The se¬
cond is that this difference arises because whereas LI learners make use of
their language faculty, adult L2 learners resort to general learning strategies.
Not surprisingly, advocates of the no-access position place considerable em¬
phasis on identifying differences between LI and adult L2 acquisition. Clah¬
sen and Muysken (1986), for instance, compare the acquisitional sequences
of German word order in LI and L2 acquisition and find ‘essential differ¬
ences’. They argue that these reflect the existence of ‘learning capacities spe¬
cific to language’ in the case of children and ‘acquisition strategies which may
be derived from principles of information processing and general problem
solving strategies’ (1986: 111) in the case of adults. In effect, Clahsen and
Muysken are arguing that where LI acquisition requires a linguistic theory,
L2 acquisition requires a cognitive theory of the kind provided by the Multi¬
dimensional Model (see Chapter 9).
The partial-access position draws on the distinction between principles
that have parameters and those that do not. Schachter (1988) raises the inter¬
esting possibility that learners may have access to linguistic principles but not
to the full range of parametric variation. This view makes two assumptions.
One is that adult learners will not manifest ‘wild grammars’ (i.e. they will not
produce ‘impossible errors’) because they are constrained by UG principles.
The other is that they will not be able to acquire the L2 values of parameters
when these differ from the LI. Clahsen and Muysken (1989), in a later paper
to that referred to above, also adopt this position, arguing that adult L2 gram¬
mars are constrained by those principles that hold for all languages (such as
structure dependency) but do not have access to parametric options. In ac¬
cordance with their earlier views, however, Clahsen and Muysken claim that
learners are not necessarily stuck with LI parameter values, as they can de¬
velop alternative hypotheses by means of general learning strategies. In some
cases, this may result in similarities between LI and L2 acquisition sequences.
Finally, Felix (1985) has advanced a dual access position. According to his
 Linguistic universals and second language acquisition 455

Competition Model (not to be confused by the model of the same name dis¬
cussed in Chapter 9), adults have continued access to UG but also make use of
‘a general problem solving module’, which competes with the language-
specific system. Felix claims that the problem-solving system is ‘a fundament¬
ally inadequate tool to process structures beyond a certain elementary level’
(1985: 51) and that this accounts for why adults fail to attain native-speaker
levels of competence. Thus, when learners reach the Piagetian stage of formal
operations at the onset of puberty, they develop the ability to form hypoth¬
eses about abstract phenomena. They are now able to call on two distinct
and, in Felix’s view, autonomous cognitive systems to deal with abstract lin¬
guistic information. Adult learners are unable to suppress the operation of the
problem-solving module. This ‘interferes’ with UG, which alone is capable of
ensuring complete grammatical competence.
Flow can we evaluate these different positions? As we have already seen, it
is doubtful whether the empirical studies of the kind considered in the previ¬
ous section provide an adequate basis for evaluation. White (1990) claims
that these studies ‘at the very least... indicate that there is accessibility via the
LI’ (1990: 131) and thus supports a partial access position. However, before
such a conclusion can be reached, it must be shown that the LI effects evident
in L2 acquisition are explicable only in terms of a UG framework and not in
terms of some other cognitive framework. For example, the fact that L2
learners appear to make initial use of their LI pro-drop setting does not con¬
stitute convincing evidence of Ll-mediated access to UG, as it can be pre¬
dicted on the basis of a more traditional model of language transfer. To be
convincing, evidence is needed that L2 learners access the cluster of features
entailed in the pro-drop parameter, but as we saw, such evidence is, at best,
weak. The available research does not produce a clear answer.
We are left, therefore, with indirect evidence and theoretical arguments.
Table 10.3 summarizes the different positions on UG access and indicates the
main assumptions of each position. The assumption that there is no critical
period for the acquisition of L2 syntax is perhaps the most questionable, as
the available evidence on the age issue (considered in Chapter 11) indicates
that adults rarely if ever achieve native-speaker levels of competence. This
calls into question the ‘complete access’ position. The ‘no access’ and ‘partial
access’ positions share two related assumptions, namely that there is a critical
period beyond which full grammatical competence is unobtainable and that
L2 is not the same as LI acquisition. Both assumptions are tenable, although
it does not follow that differences between LI and L2 acquisition are the res¬
ult of lost or diminished access to UG, as they may reflect other variables, such
as general cognitive development and socio-affective factors. The ‘no access’
and ‘partial access’ positions are distinguished in terms of their assumptions
regarding ‘wild grammars’. This is clearly an aspect of acquisition that needs
further study and it is probably premature to reach any firm conclusion. It is
difficult, therefore, to choose between the ‘no access’ and ‘partial access’
456 Internal factors

positions. It is also difficult to evaluate the dual access position. The key as¬
sumption is that adults will manifest types of linguistic behaviour not seen in
children. The problem here is that it is extremely difficult to demonstrate that
differences are attributable to the operation of different learning systems, for,
as Cook (1985) has pointed out, the enhanced channel capacity of the adult
learner, evident in ‘development’, can mask similarities in ‘acquisition’.
It is clear from this discussion that no verdict can be reached. White
(1989a) claims there is a growing consensus in favour of the view that UG is
available via the LI, but neither the empirical evidence currently available nor
theoretical arguments provide uncontrovertible support. In particular, it is
not yet clear that the LI effects that have been identified require a UG
explanation.

Position Description Main assumptions

Complete LI provides learners with a ‘quick’ setting L2 learners will be able to

access for the L2 parameter if the value is the
same, otherwise, the L2 learner proceeds
in same way as the LI learner. L2 learners
have full access to UG principles.
No access (the L2 learners no longer have access to the
Fundamental principles and parameters of UG; general
Difference learning strategies replace UG.
Hypothesis)
Partial access L2 learners have full access to UG
(i.e. via LI) principles but can only access those
parameters operative in their LI; they may
be able to reset LI parameters by means
of general learning strategies.
Dual access (the L2 learners have access to UG but this is
Competition partly blocked by the use of general
Model) learning strategies.
attain full linguistic
competence; there is no
critical period blocking L2
acquisition.
L2 * LI acquisition; adults
fail to achieve full linguistic
competence; ‘wild
grammars’ can occur.
L2 and LI acquisition are
the same in part; adults fail
to achieve full linguistic
competence; no ‘wild
grammars’ are evident.
L2 = LI acquisition in part;
adults fail to achieve full
linguistic competence;
adults manifest similar and
different linguistic
behaviour to children.

Table 10.3: Alternative positions regarding access toUG in L2 acquisition

The role of negative evidence

A third issue of considerable theoretical importance is the role of negative

evidence in L2 acquisition. We have already seen that this is typically not
available to the LI learner, whereas the adult L2 learner may have access to
both corrective feedback and to explicit grammatical information. The gen¬
eral effect that these have on L2 acquisition is considered in Chapter 14. Here
we will consider negative evidence in relation to UG.
 Linguistic universals and second language acquisition 457

Assuming for the moment that the provision of negative evidence is indeed
beneficial to L2 learning, two theoretical positions are tenable. One is that
negative evidence enables learners to acquire grammatical properties that
would otherwise be lost because they do not have continued access to learn¬
ing principles. This is the view that White (1991) adopts. She provides evid¬
ence to show that adverbial placement rules in L2 English, which are typically
not acquired by learners whose LI lacks them (see the earlier section relating
to the Subset Principle, page 436), can be successfully learnt through formal
instruction. The claim is that negative evidence triggers the resetting of a
parameter to its L2 value.
The second position, associated with Schwartz (for example, Schwartz
1986 and Schwartz and Gubala-Ryzak 1992), is that UG can be activated
only by means of positive evidence and that negative evidence, therefore,
plays no role in UG-based acquisition. Schwartz acknowledges that negative
evidence can result in the acquisition of grammatical knowledge, but argues
that there is no mechanism that can ‘translate’ this knowledge into input of
the type required by UG. In support of this position, Schwartz and Gubala-
Ryzak reanalyse the data from White (1991) in order to argue that, while the
learners were clearly successful in temporarily eliminating an incorrect adver¬
bial placement rule, they achieved this without restructuring their in¬
terlanguage grammars. This is tantamount to claiming that negative evidence
aids the development of explicit L2 knowledge, but not implicit (see Chapter
9).
Again, the resolution of this argument requires us to distinguish between
the operation of UG and general learning strategies, which, as we have
already noted, is no easy task. We might note, however, that if UG exists to
enable children to acquire grammatical competence solely on the basis
of positive evidence, as is generally accepted, it is hardly felicitous to
propose that L2 learners can access parts of it with the help of negative
evidence.

Universal Grammar and second language acquisition: an evaluation

Some researchers are strongly committed to a UG-based theory of L2 acquisi¬
tion, while others are equally strongly opposed. We will begin this evaluation
by considering some of the objections that have been levelled at the theory.
These objections concern (1) the domain of the theory, (2) methodological
issues, (3) the empirical research based on the theory, and (4) the principal
tenets of the theory. We will conclude with a consideration of the theory’s
strengths.
458 Internal factors

The domain of the theory

Earlier, we saw that a UG-based theory of L2 acquisition is directed at ex¬
plaining a fairly restricted phenomenon—that part of grammatical compet¬
ence that is determined by an innately specified and abstract knowledge of
grammatical principles. Much, therefore, is excluded. The theory does not
address how learners acquire the ‘skill’ of using their grammatical know¬
ledge, and it also ignores other aspects of competence—how learners develop
their lexicon, how they construct form-meaning networks, and how they
learn to perform speech acts appropriately, for example. Indeed, the theory
does not even fully explain how ‘grammar’ is acquired, as it concerns itself
only with those aspects of grammar that fall within the ‘core’. In the eyes of
some researchers, the theory is unconvincing because it fails to explain crucial
aspects of language and language use.20

Methodological problems
A number of methodological problems have already been considered, for ex¬
ample, the over-reliance on grammaticality judgement tests and the relative
lack of longitudinal studies (see page 441). A further problem concerns the
definition of ‘adult’. This is of considerable importance, as a UG-based theory
of L2 acquisition is, in the main, a theory of adult language acquisition. Child
L2 learners are assumed to have the same access to UG as LI learners.21 The
key issue is whether adult L2 learners are also guided by UG. To examine this
issue it is necessary to investigate learners who started to learn an L2 in adult¬
hood, but many of the studies do not do this. It is also necessary to determine
when ‘adulthood’ commences. As we will see in Chapter 11, the critical age
for grammar appears to be fairly late, around 15 years. Few of the studies to
date have examined learners who began their L2 learning after this age.
Methodological problems of these kinds do not in themselves pose major
objections to the theory, however. The answer lies in better designed re¬
search, a point fully recognized by UG-oriented researchers.

The indeterminacy of the results obtained from empirical research

Given these methodological problems, it is, perhaps, not surprising to find
that the empirical research has produced such indeterminate results. We ex¬
amined studies directed at three issues—the availability of UG in adult L2 ac¬
quisition, the role of markedness, and learners’ access to learning principles
such as the Subset Principle. While the research did suggest that learners have
at least partial access to UG, more specific conclusions were not possible.
Again, this need not be seen as a serious defect, for as White (1989a) points
out ‘what we are witnessing here is developing theories being applied to a par¬
ticular domain’ (1989: 137). In such a case, indeterminacy can be expected.
 Linguistic universal and second language acquisition 459

Theoretical problems

Theoretical problems are evident in both the theory of language and the
model of grammar that together inform L2 theory. As we have seen, the de¬
tails of the model of grammar are constantly changing. How many settings
does a particular parameter have? What does each setting consist of? Which
grammatical features cluster within a parameter? Not surprisingly, such
questions have produced a variety of answers as linguists explore the various
possibilities. The continual revision of the model is problematic to L2 re¬
searchers, however, who may find themselves working with a model that be¬
comes obsolete before they can complete their research.
A more serious problem, though, is that of falsifiability. How can we be
sure, for instance, that learners are or are not behaving in accordance with
UG? Learners’ behaviour is not categorical, but variable, or as Hilles (1986:
234) calls it, ‘fuzzy’. Can this be dismissed as performance variability, or does
it reflect an indeterminate competence—‘a period during which the para¬
meters may waver between two values’? The central issue is what constitutes
confirming and disconfirming evidence of hypotheses based on UG theory.
Do learners demonstrate access to UG if they perform judgements correctly
above the level of chance, or if they produce sentences providing evidence of
parameter resetting just 5 per cent (or even 1 per cent) of the time? Con¬
versely, do they demonstrate lack of access to UG if they perform judgements
incorrectly above the level of chance or provide evidence in the sentences they
produce of ‘impossible errors’? Bley-Vroman et al. (1988) note:

Universal Grammar, after all, is concerned with the notion of possible

grammar. The consequences of UG are not merely statistical. Violations ...
are not just ungrammatical ‘more often than not’. They are ungrammat¬
ical—period (1988: 27).

And yet, most researchers have examined the ‘statistical consequences’. One
possible solution can be found in Uziel’s (1993: 55) proposal that researchers
abandon testing whether there is ‘absolute obedience’ to a given principle,
and instead focus attention on the ‘relative acceptability of certain syntactic
constructions as it is predicted by the principles of UG’.
The falsifiability problem is evident in two other ways. It is generally ac¬
cepted, even by UG theorists, that adult L2 learners can make use of faculties
other than their language faculty to learn L2 rules. How, then, can we distin¬
guish between UG-based and general cognitive learning? There is a pressing
need for researchers to agree on empirically verifiable indicators of these two
types of learning, so as to improve their ability to test UG-inspired hypoth¬
eses. Similarly, a number of researchers (for example, Bley-Vroman et al.
1988; Schachter and Yip 1990) have pointed out that the difficulties which
many learners experience with sentences that violate UG principles may
460 Internal factors

derive from problems with parsing rather than from their inability to access
UG. Again, how can we distinguish these?
The validity of the ‘logical problem of language acquisition’, as stated by
Chomsky and others, has also been challenged. Klein (1991), for example,
disputes the existence of this problem on the grounds that (1) LI acquisition is
not instantaneous (it is ‘a difficult and cumbersome process that extends over
many years’), (2) it is an ‘essentially accumulative process’, and (3) ‘it presup¬
poses a vast amount of input’ that obviates the need for a specific language
faculty. On the basis of these claims (and others of a less controversial na¬
ture), Klein challenges the assumption that a theory of language acquisition
needs to posit the existence of a special language faculty. He suggests that the
assumption of such a faculty runs up against Occam’s razor—the require¬
ment that no theory should contain unnecessary principles. He comments:

A theory that can do without the assumption of a specific ‘language mod¬

ule’ is much better than a theory which requires the assumption ... If every¬
thing can be explained without an extra cognitive capacity, why assume it?

Hyams (1991) tackles each of Klein’s claims head on. She points out: (1)
that children are able to learn certain complex structures with ‘remarkably
little effort’ (1991: 73), (2) that‘the assumption that acquisition is cumulative
in no way eliminates the logical problem inherent in the instantaneous model’
(1991: 76), and (3) that even if large amounts of input are required by the
child (a point she doubts), this does not justify the claim that input is more
than just a trigger of the language acquisition device.
It is difficult to evaluate the claims that children learn with little or consid¬
erable effort, as neither Klein nor Hyams explain what they mean by these
epithets. The essential points are (2) and (3), however. Point (2) rests on evid¬
ence provided by linguists that certain syntactical constructions (like question
formation in English) are not learnable on the basis of input alone. O’Grady
(1991) and Parker (1989), however, have presented linguistic arguments to
demonstrate that certain types of knowledge, such as those found in the Sub-
jacency Principle and the pro-drop parameter, are not as ‘abstract’ as Chom¬
sky and others have claimed, and can be accounted for in terms of input.
O’Grady, for instance, provides an interesting, non-UG based explanation
for pro-drop phenomena in the LI and L2 acquisition of English. He suggests
that initially learners are unable to distinguish between finite and non-finite
verbs and thus are unable to distinguish between subject-taking and non¬
subject-taking verbs. He presents evidence to show that subjects emerge in the
speech of LI learners when they acquire tense. O’Grady’s position is similar
to Klein’s, namely that ‘there is an acquisition device that is especially well
suited for the task of grammar instruction, but it does not include “UG” or
any other inborn syntactic knowledge’ (1991: 340). He argues that learners
can make much fuller use of ‘positive evidence’ than Chomsky has allowed
for. Although it is clearly premature to reject the linguistic arguments that
 Linguistic universals and second language acquisition 461

have been advanced in support of an innate language faculty, it is equally

clear that alternative arguments may become available.
It is also difficult to arrive at firm conclusions regarding (3). Hyams is obvi¬
ously right to emphasize that the issue of quantity of input is irrelevant where
its function is concerned. Irrespective of whether input is abundant or not, if it
can be shown that input functions as a ‘trigger’, the UG position is supported.
However, the evidence for a ‘triggering’ effect (for example, in the case of the
pro-drop parameter) is mixed, and is sometimes confounded by alternative
accounts of the relevant parameters. Hyams’ other argument against the im¬
portance of input is that learners deprived of large amounts of input (for ex¬
ample, the children of deaf parents) show no delay in language development.
However, this cannot be considered evidence in support of a specific language
faculty. It demonstrates only that children have a remarkable capacity for
language learning, which may be specifically linguistic or more generally cog¬
nitive in nature.
The Klein-Hyams debate mirrors an ongoing controversy in language ac¬
quisition research, namely whether it is necessary to posit a special language
faculty. According to UG theory it is; according to the cognitive theories ex¬
amined in the previous chapter it is not. At the moment, it is difficult to see
whether the argument has been conclusively won one way or the other.

The strengths of a UG-based theory of second language acquisition

Despite the controversy surrounding the idea of a special language faculty,
there is general acceptance that at least some aspects of language learning
concern purely formal properties of language. It would follow that it is not al¬
ways essential to investigate form-meaning relationships (see Foster 1990
for a development of this argument). UG provides a theoretical basis for an
examination of the way L2 learners acquire purely formal properties
of language.
UG-inspired SLA research enjoys one obvious advantage. In contrast to
many of the cognitive theories of L2 acquisition we have considered in this
chapter, UG theory affords very precise hypotheses about specific linguistic
properties. Also, in contrast to many other theories, UG-based theory is very
strictly defined, covering only part of the total phenomena which a compre¬
hensive theory of L2 acquisition will need to account for. While in the eyes of
some this constitutes a limitation, in the eyes of many it is an enormous
strength, as it provides a means for delimiting the field of enquiry to manage¬
able proportions. It is also possible that L2 research can contribute to the de¬
velopment of linguistic theory. UG-based theory affords a way of unifying the
efforts of linguists and SLA acquisition researchers.
462 Internal factors

Conclusion
Any explanation of L2 acquisition must take account of what learners are try¬
ing to learn—language. The explanations considered in this chapter have
sought to achieve this by drawing on the findings of two kinds of linguistic
enquiry—the study of language typology and of Universal Grammar. We
have seen that the former, which involves the crosslinguistic description of
languages with a view to identifying universal, has afforded a number of
insights into how learners acquire an L2 but is unable to provide an adequate
explanation because the linguistic universals it has identified have no theoret¬
ical status. In contrast, we have seen that Universal Grammar offers a power¬
ful theoretical account of L2 acquisition, but, to date, lacks convincing
empirical support for many of its specific hypotheses and also faces a number
of methodological problems, although in this respect it is arguably in no
worse shape than other theories.
We should also note that both approaches are directed at explaining only a
subset of the total L2 phenomena that must be ultimately be accounted for.
They have dealt exclusively with a limited set of grammatical phenomena.
This is both a limitation and a strength of linguistic theories of L2 acquisition.

Notes

1 Chomsky writes: ‘I have not hesitated to propose a general principle of

linguistic structure on the basis of observations of a single language’
(1980a: 48).
2 A limited number of adverbials (for example, ‘never’, ‘hardly’, ‘scarcely’)
require subject-verb inversion in English when they occur sentence-
mitially. However, most adverbials do not. In contrast, all sentence-initial
adverbials require inversion in German.
3 Croft identifies a fourth factor involved in markedness— ‘neutral value’
(for example, ‘old’ is used in both statements and in neutral questions re¬
ferring to someone’s age)—but argues that this is of less importance in
typological study.
4 Evidence for the claim that learners position the negator in relation to the
‘main verb’ rather than the ‘finite element’ can be found in the fact that
initially learners produce sentences like:

han kan inte komma (he can not come)

which are grammatically correct. It is counter-intuitive, however, to ar¬

gue that learners acquire correct negator position in verb phrases with an
auxiliary before those consisting simply of main verbs. As Jordens points
out it is more plausible to argue that negative placement is determined ini¬
tially by the learners’ recognition of what constitutes the main verb. The
early rule, negator + mam verb, accounts for negator position in all
sentences.
 Linguistic universals and second language acquisition 463

5 Cook (1988:170) distinguishes a ‘general level of the theory’ (i.e. the The¬
ory of Universal Grammar), which has changed little from its inception,
from ‘the most particular’ (i.e. the Government/Binding Model), which
changes almost from minute to minute.
6 It is, in fact, arguable whether English is a non-pro-drop language, as we
saw in Chapter 6 when discussing the study by Gass and Lakshmanan
(1991). This study showed that speakers of standard English regularly
drop pronouns in the speech they address to learners. Also, certain dia¬
lects and registers of English permit the pronoun to be dropped.
7 Linguists do not always agree about what features cluster in a particular
parameter or indeed whether parameters entail clustering. Thus, Chom¬
sky and Hyams offer somewhat different accounts of the pro-drop para¬
meter (see page 431), while Wexler and Mancini (1987) present an
account of parameters that does not allow for clustering.
8 If Felix’s proposal is correct, ‘wild grammars’ may be possible. That is,
children may construct grammars that conform only with those prin¬
ciples of UG that are available to them at a given time and do not conform
to those principles which have not yet maturated.
9 There are, in fact, two different proposals regarding how children come
to construct conservative grammars (see White 1989b: 139) for a discus¬
sion and further references). One concerns the Subset Principle, as de¬
scribed in the text. This is seen as extraneous to UG, part of a separate
module of the language faculty. The other suggests that parameters are
ordered according to markedness, such that children automatically opt
for an unmarked before a marked parameter and by so doing avoid build¬
ing grammars that cannot be subsequently disconfirmed. This proposal
obviates the need for a separate Subset Principle, if it can also be shown
that all subset problems are handled by means of parameters.
10 Generative descriptions focus on the formal properties of language on the
basis that these can be separated from functional properties. The prin¬
ciples and parameters of UG, for example, relate exclusively to formal
properties. Chomsky has argued consistently against the view that lan¬
guage should be viewed as a tool of communication, pointing out that it is
often used for non-communicative purposes (such as thinking). His views
contrast starkly with those of functional grammarians like Givon (1979)
and Halliday (1978), who treat form and function as inseparably linked.
11 Measuring the time it takes for learners to make judgements may provide
a way of determining what kind of knowledge they use. Bialystok’s
(1979) study indicated that learners first try to use their implicit know¬
ledge, and only when this fails to provide them with a clear answer do
they resort to explicit knowledge. Very few studies, however, have used
timed judgements (see Cook 1990 for an exception).
12 Eubank (1987a) also argues that failure in parsing can account for the dif¬
ficulties learners experience with pronominal reference in sentences like
464 Internal factors

‘When he entered the office, the janitor questioned the man’. Eubank
points out that ‘any study that considers performance data must take into
account multiple sources of variation’ and must be prepared to consider
‘multiple cognitive mechanisms’ (1987a: 63). In other words, Eubank
queries whether the results obtained from Uij-inspired studies can be
interpreted solely in terms of UG.
13 As is evident from this short survey of studies investigating the availability
of the Subjacency Principle, the linguistic theory on which they are based
has changed. In addition to the differences between the GovernmentI
Binding and Barriers models, other problems arise. White (1989a), for
example, cites Saito (1985) as arguing that Subjacency does exist in lan¬
guages like Japanese. If this is the case, the assumption of several of the
studies (for example, Ritchie and Bley-Vroman et al.) that Japanese and
Korean learners do not have access to Subjacency in ^-questions via
their LI is not justified.
14 The data from White’s question-formation task lent support to the results
obtained from the grammaticality judgement task. The Spanish subjects
were more likely to produce a ‘that’-trace in tt^-questions than the
French, which White saw as further evidence of the influence of the LI
parameter setting.
15 Gass (1989) cites a study by Lakshmanan (1986) which used grammat¬
icality judgements to investigate Spanish-, Arabic-, and Japanese-speak¬
ing learners of L2 English. This study also indicated that L2 learners make
use of their LI parameter setting but not with all its hypothesized
properties.
16 Hilles’ (1991) study also addressed a second question—whether the ab¬
sence of pronominal subjects in the learners’ early interlanguage was the
result of the transfer of the LI setting or the result of treating English as a
topic-deleting language. The results were not conclusive, however.
17 Lakshmanan offers a number of interesting explanations for her findings.
For instance, she suggests that Uguisu used subject pronouns from the
start because English, unlike Japanese, does not permit null subjects of
any kind. In other words, her hypothesis might have been ‘If no null sub¬
jects, then no null pronouns’. This, of course, has nothing to do with UG.
18 White argues that adverb placement is related to a binary parameter of
UG, the Adjacency Condition on Case Assignment, which requires that
an NP with case must be next to its case assigner.
19 In the case of L2 acquisition, parameter setting is frequently ‘parameter
resetting’. It should be noted that the idea of ‘parameter resetting’ is not
restricted to proponents of complete access. Theorists like White, who
adopt the view that learners can only access UG via their LI, also allow
for parameter resetting with the help of negative evidence.
20 What is deemed the ‘crucial’ domain of a theory depends on the purpose
for which the theory is intended. This point is developed in Chapter 15.
 Linguistic universals and second language acquisition 465

21 The assumption that child L2 learners acquire grammar in the same way
as LI learners needs to be tested, of course. Some studies (such as Hilles
1991)have done so.

Further reading
There are two helpful overview articles on linguistic universals and L2
acquisition:
P. Lightbown and L. White, ‘The influence of linguistic theories on language
acquisition research: description and explanation.’ Language Learning 37
(1988): 483-510.
S. Gass, ‘Language universals and second-language acquisition.’ Language
Learning 39: (1986) 497-534.
A good starting point for reading about typological universals and second
language acquisition is one of the general survey books on language typology
such as
B. Comrie, Language Universals and Linguistic Typology (Basil Blackwell,
1981)or
W. Croft, Typology and Universals (Cambridge University Press, 1990).
The main book for the study of typological universals as they relate to L2 ac¬
quisition is
W. Rutherford (ed.): Language Universals and Second Language Acquisition
(John Benjamins, 1984).
Interesting empirical studies based on typological universals include:
K. Hyltenstam, ‘The use of typological markedness conditions as predictors
in second language acquisition: the case of pronominal copies in relative
clauses’ in R. Andersen (ed.): Second Languages: A Cross-linguistic Perspect¬
ive (Newbury House, 1984).
R. Hawkins, ‘Do second language learners acquire restrictive relative clauses
on the basis of relational or configurational information? The acquisition of
French subject, direct object and genitive restrictive relative clauses by second
language learners.’ Second Language Research 5 (1989): 158-88.
A very readable account of Chomsky’s ideas about language can be found in
V. Cook, Chomsky’s Universal Grammar (Basil Blackwell, 1988).
L. White, Universal Grammar and Second Language Acquisition (John Ben¬
jamins, 1989) provides an excellent account of UG-based work in SLA
research.
There are also a number of collections of papers dealing with UG and L2 ac¬
quisition, of which the following are perhaps the most significant:
S. Flynn and W. O’Neill (eds.): Linguistic Theory in Second Language Ac¬
quisition (Kluwer, 1988).
S. Gass and J. Schachter (eds.): Linguistic Perspectives on Second Language
Acquisition (Cambridge University Press, 1989).
466 Internal factors

L. Eubank (ed.): Point Counterpoint: Universal Grammar in the Second Lan¬

guage (John Benjamins, 1991).
Journal articles on UG and L2 acquisition can be found in abundance, in par¬
ticular in Second Language Research and Language Learning. It is difficult to
pick out ‘key’ articles, but the following raise important issues:
L. White, ‘The “pro-drop” parameter in adult second language acquisition.’
Language Learning (1985) 35 : 47-61.
S. Felix, ‘More evidence on competing cognitive systems.’ Second Language
Research (1984) 1: 47-72.
H. Clahsen and P. Muysken, ‘The availability of Universal Grammar to adult
and child learners - a study of the acquisition of German word order.’ Second
Language Research (1986) 2: 93-119.
J. Schachter, ‘Second language acquisition and its relationship to Universal
Grammar’. Applied Linguistics (1988) 9: 219-35.
R. Bley-Vroman, S. Felix, and G. Ioup, ‘The accessibility of Universal Gram¬
mar in adult language learning.’ Second Language Research (1988) 4: 1-32.
S. Uziel, ‘Resetting Universal Grammar parameters: evidence from second
language acquisition of Subjacency and the Empty Category Principle.’ Se¬
cond Language Research (1993) 9: 49-83.
PART FIVE

Explaining individual differences

second language acquisition
Introduction

The chapters in the previous section of the book treated second language (L2)
acquisition as a phenomenon that has universal, structural properties, and
advanced various explanations for them. The underlying assumption, based
on the observed regularities in learner language described in Part Two, was
that all learners analyse input and store information about the L2 in much the
same way. However, it is also true that learners vary enormously in both the
ways they set about learning an L2 and also in what they actually succeed in
learning. The study of individual learner differences (IDs) comprises an im¬
portant area of work in SLA research and contributes to theory development.
The study of IDs in SLA research seeks answers to four basic questions: (1)
In what ways do language learners differ? (2) What effects do these differ¬
ences have on learning outcomes? (3) How do learner differences affect the
process of L2 acquisition? and (4) How do individual learner factors interact
with instruction in determining learning outcomes? We will consider the first
three questions in this part of the book. Chapter 11 will address questions (1)
and (2). Here we will describe various kinds of individual differences relating
to learners’ beliefs, their affective states, and certain key general factors such
as age, language aptitude, and motivation. These differences help us to under¬
stand why learners vary in how quickly they learn and why most learners fail
to achieve native-speaker competence. Chapter 12 will focus on question (3)
by addressing the learning strategies learners use to obtain and understand in¬
put, to retain new L2 information, and to regulate their own language learn¬
ing. Answers to question (4) will be delayed until Chapter 14, when we
consider the role of formal instruction in L2 acquisition.
We have already examined a number of factors that distinguish learners. In
Chapter 6 we studied various social factors (in particular social class, sex, and
ethnicity) and saw that these influence the extent and type of exposure to L2
input. These factors affected learners as groups—hence the label ‘social’. In
this section, we focus on factors that affect learners as individuals and that are
psychological in nature. A full account of how learners differ with regard to
how, how much, and how fast they learn a L2 will need to take account of
both social and psychological factors, and how these interact.
■
11 Individual learner differences

Introduction

There is a veritable plethora of individual learner variables which researchers

have identified as influencing learning outcomes. Table 11.1 lists the main
variables mentioned in three surveys. It demonstrates the importance at¬
tached to individual differences (IDs) by different researchers, and also the
different ways they classify them. The constructs referred to in Table 11.1 are
often vague and overlap in indeterminate ways. This makes it difficult to
synthesize the results of different studies, and even more difficult to arrive at a
coherent overall picture.
Another problem, related to that of classifying learner variables, is the
choice of terms for labelling the different factors. Often there is no clear dis¬
tinction drawn in the use of terms like ‘belief’, ‘attitude’, ‘state’, and ‘factor’,
while often it is not clear what terms likes ‘learning style’ and ‘motivation’ re¬
fer to. Further, as we will see in Chapter 12, the term ‘strategy’ has been used
to refer to such an assortment of notions that it is in danger of becoming
vacuous.
Despite these problems the study of IDs has attracted a lot of attention in
SLA research, and has made considerable advances. As Table 11.1 shows,
there are a number of dimensions of learner differences which are generally
acknowledged (for example, age, aptitude, motivation, cognitive style, and
learning strategies).
The chapter begins by attempting to define a number of central constructs
and by outlining a framework for examining IDs. Next, it considers the main
ways in which IDs have been studied. The bulk of the chapter is taken up with
surveys of research which has investigated learners’ beliefs about language
learning, their affective states, and various general factors (‘age’, ‘aptitude’,
‘learning style’, ‘motivation’, and ‘personality’). No attempt will be made,
however, to examine all the dimensions of inter-learner variation listed in
Table 11.1. This is because little is known about many of these dimensions
and because the IDs mentioned above have been considered the ones of great¬
est importance for L2 acquisition.
472 Explaining individual differences

Altman (1980) Skehan(1989) Larsen-Freeman and Long

(1991)

1 Age 1 Language aptitude 1 Age

2 Sex 2 Motivation 2 Socio-psychological
3 Previous experience with 3 Language learning factors
language learning strategies a motivation
4 Proficiency in the native 4 Cognitive and effective b attitude
language factors 3 Personality
5 Personality factors a extroversion/introversion a self-esteem
6 Language aptitude b risk-taking b extroversion
7 Attitudes and motivation c intelligence c anxiety
8 General intelligence (IQ) d field independence d risk-taking
9 Sense modality e anxiety e sensitivity to rejection
preference f empathy
10 Sociological preference g inhibition
(e.g. learning with peers h tolerance of ambiguity
vs. learning with the 4 Cognitive style
teacher) a field independence/
11 Cognitive styles dependence
12 Learner strategies b category width
c reflexivity/impulsivity
d aural/visual
e analytic/gestalt
5 Hemisphere specialization
6 Learning strategies
7 Other factors e.g. memory,
sex

Table 11.1: Factors listed as influencing individual learner differences in language

learning in three surveys

A framework for investigating individual learner differences

The framework that will guide our examination of IDs is shown in Figure
11.1. Three sets of interrelating variables are identified. The first set consists
of IDs, which are of three main types. Learners have been shown to have be¬
liefs about language learning. Horwitz (1987a) and Wenden (1987a) have
shown that learners have strong, pre-conceived ideas about such issues as the
importance of language aptitude, the nature of language learning, and the
strategies that are likely to work best. Second, learners have been shown to be
strongly influenced by their affective states (see Bailey 1983). Some learners
are fearful of starting to learn an L2, while some are confident. Some develop
anxiety as a result of their competitive natures and their perceptions of
whether they are progressing or not. Both learners’ attitudes and their af¬
fective states are subject to change as a result of experience. Third, there are
various general factors. These constitute major areas of influence on learning
and can be ranged along a continuum according to how mutable they are. For
 Individual learner differences 473

example, language aptitude is generally considered a stable factor, not readily

influenced by the environment (Carroll 1981), while certain types of motiva¬
tion are likely to change as a result of the learner’s learning experiences (see
Berwick and Ross 1989; Crookes and Schmidt 1990). The general factors
also vary according to the extent of the learners’ control over them. For ex¬
ample, learners can do nothing about their age, but they may be able to
change their learning style (Thomas and Harri-Augstein 1990). Clearly,
beliefs, affective states, and general factors are interrelated. For example,
learners’ beliefs and their affective responses to learning situations may
be influenced by personality variables. One of the goals of ID research is
to identify the nature of these interrelationships.
The second set of variables consists of the different strategies that a learner
employs to learn the L2. These will be considered in detail in the following
chapter, together with studies of the ‘good language learner’. The third set
concerns language learning outcomes. These can be considered in terms of
overall L2 proficiency, achievement with regard to L2 performance on a par¬
ticular task, and rate of acquisition. Learning outcomes constitute the ‘prod¬
ucts’ of the acquisitional process.
(1)
Individual learner differences

- beliefs about language learning

- affective states
- general factors

outcomes
- on proficiency
- on achievement
- on rate of
acquisition
Figure 11.1: A framework for investigating individual learner differences

These three sets of factors are related in complex ways. ID research to date
has concentrated on investigating the effects of different ID variables on
learner proficiency, achievement, or rate of progress, measured in terms of
performance on some kind of language test. There has also been little work
done on the effect that learning outcomes can have on IDs. It is likely that the
more mutable factors exist in a symbiotic relationship with achievement.
For example, anxiety and motivation influence learning (positively or negat¬
ively), but perceived success or failure in learning will also have an effect on
474 Explaining individual differences

motivation. The strategies learners employ will be influenced by ID variables

and may also have an effect on them, for instance, when successful use of a
particular strategy leads to enhanced motivation or reduced anxiety. Finally,
as we will see in Chapter 12, strategy choice can affect learning outcomes,
which in turn can have an influence on strategy use (for example, certain stra¬
tegies may only become available when learners have reached a threshold of
L2 proficiency).
In the inner part of the triangle in Figure 11.1 are ‘learning processes and
mechanisms’, so positioned because they are largely hidden. These account
for how input becomes intake and how intake is integrated into the learner’s
interlanguage system; they were considered in Part Four of this book. The ex¬
tent to which learning processes are influenced by IDs or by conscious use of
learner strategies is a matter of some dispute. Seliger (1984) has claimed that
IDs do not affect the ‘deep’ processing involved in L2 acquisition. To date,
there has been little attempt to examine the relationship between IDs and the
process of acquisition by examining whether different kinds of learners pro¬
duce different kinds of errors or manifest different developmental patterns.

The methodology of individual difference research

Following Skehan (1989), it is possible to distinguish two general approaches
to the study of IDs: (1) the hierarchical approach, and (2) the concatenative
approach. These concern the relationship between theory and research.

The hierarchical approach

The hierarchical approach has as its starting point a theory that affords pre¬
dictions about how particular IDs affect learning. These predictions generally
take the form of specific hypotheses which can be tested empirically (i.e. they
can be confirmed or disconfirmed by carrying out studies that have been spe¬
cially designed to investigate them). As Skehan notes, however, this approach
has been relatively little used in ID research because of the absence of a suffi¬
ciently detailed theory of IDs.

The concatenative approach

This is a research-then-theory approach. Its starting point is the identification
of a general research question (such as ‘To what extent does motivation ac¬
count for L2 achievement?’). The next step is to collect data that can be used
to identify various IDs (the independent variables), and perhaps also to ob¬
tain measures of L2 learning (the dependent variable). Various procedures
are then used to investigate the nature of the relationships among the inde¬
pendent variables (for example, to see whether ‘aptitude’ is related to ‘mo¬
tivation’) and also the relationship between the independent variable(s) and
 Individual learner differences 475

the dependent variable. Such research is typically correlational in nature; that

is, it can demonstrate the existence of relationships but cannot easily deter¬
mine what is cause and what is effect. Nevertheless, it can contribute to the¬
ory development. The concatenative approach has been widely used in SLA
research, a good example being the study of motivation by Gardner and his
associates (see Gardner 1985).
In addition to these two general approaches, it is possible to distinguish
two general research traditions in ID research: (1) naturalistic research, and
(2) confirmatory research. These traditions are evident in all branches of the
social sciences. They concern the way subjects are chosen for study, the na¬
ture of the data that are collected, and how they are typically analysed.

Naturalistic research

In naturalistic research people are studied in real-life settings. There is no at¬

tempt to manipulate the learning context, nor are subjects randomly assigned
to pre-determined groups. Naturalistic research also typically involves the
collection of qualitative data.
The major method for collecting data in naturalistic research is the detailed
observation of subjects and settings. However, this method has not been very
successful in the study of IDs, as what learners do often does not reveal much
about their psychological states and characteristics or the strategies they use
to learn (see Rubin 1975 and Cohen 1984). However, methods involving in¬
trospection and retrospection have proved helpful. Diary studies have pro¬
vided a wealth of information about IDs (see Fry 1988; Bailey 1991). In these,
learners are asked to keep journals of their progress in learning an L2. The
journals are then subjected to careful analysis in order to identify the signific¬
ant variables and how they influenced learning. Learners can also be asked to
think aloud as they perform typical language-learning tasks and the resulting
protocols subjected to careful analysis, or they can be interviewed about their
beliefs, attitudes, reactions, and strategies. It should be pointed out, though,
that the validity and reliability of these kinds of self-report methods is not uni¬
versally accepted. (See the papers in Fserch and Kasper (1987) for a discussion
of introspective methods.)
Another naturalistic method, widely accepted but little used in ID research,
involves the collection and analysis of samples of learner language, an ap¬
proach characteristic of the descriptive research in mainstream SLA research
(see Part Two of this book). However, whereas the mainstream research has
focused on identifying general features of learner language in order to estab¬
lish universal processes of L2 acquisition, ID research has sought to identify
differences in the learner language produced by individual learners. The gen¬
eral procedure consists of obtaining data from two or more learners per¬
forming the same or similar tasks, analysing the data in order to identify
differences, and then trying to identify different ‘performance styles’ or ‘task
476 Explaining individual differences

orientations’. Hatch’s (1974) distinction between ‘rule formers’ and ‘data

gatherers’ (see the section on Learning styles from page 499 onwards) was
based on a series of studies of individual learners which employed product
analysis.

Confirmatory research
In contrast to naturalistic enquiry, confirmatory research is interventionist in
nature. That is, it seeks to control the learning environment and to manip¬
ulate key variables. This kind of research does not typically make use of intact
groups of subjects, but rather assigns them to specific groups. Confirmatory
research is often conducted by means of carefully designed experiments
involving some kind of ‘treatment’. Data are collected by asking learners to
complete various kinds of tests. These afford numerical data which can be
analysed statistically. A good example of this kind of research can be found in
Gardner and MacIntyre’s (1991) study of instrumental motivation. This
study involved two groups of learners—an experimental and a control
group—who received different treatments regarding how they were re¬
warded for performing a vocabulary learning task. The difference between
the two groups was measured by reference to their scores on the vocabulary
task. Confirmatory research can also be correlational in nature, using data
collected from tests or closed-item questionnaires to establish whether pre¬
dicted relationships occur. Hansen and Stansfield’s (1981) study of the rela¬
tionship between learners’ cognitive styles and their performance on various
tests of linguistic and communicative competence is an example of this kind
of research.
ID research, then, can be characterized in terms of (1) the general ap¬
proach, and (2) the research tradition, affording four main types, as shown in
Figure 11.2. Concatenative research of both type B and type D has been the
principal method of enquiry, but hierarchical research in the confirmatory
tradition (type C) is gaining in popularity, as theories are developed. Hier¬
archical research that employs a naturalistic methodology (type A) is the least
well-represented.
ID researchers are increasingly making use of more than one way of col¬
lecting information about learners in a single study. In this way they can ob¬
tain a ‘rich’ databank. For example, Abraham and Vann (1987) and I (Ellis
(1989a) conducted case studies of individual learners involving a concat¬
enative approach that included introspective, performance, and test data.
Such research, which allows for triangulation (the use of diverse kinds of data
as a means of achieving more accurate and reliable results), seems particularly
promising in shedding light on the complex relationships that exist among ID
variables.
We turn now to a survey of the research which has investigated IDs, begin¬
ning with an account of learners’ beliefs about language learning.
 Individual learner differences 477

Hierarchical Concatenative
approach approach

A B

Theory-led studies Exploratory studies

Naturalistic that examine IDs that examine IDs in the
tradition in how learners learn way learners learn and
and use the L2 in use the L2 in
naturally occurring naturally occurring
settings. settings
(e.g. Hatch’s (1974)
review of descriptive
studies of L2 learners)

C D

Theory-led studies Correlational studies

that examine IDs
experimentally, i.e.
by assigning learners
Confirmatory to special groups and
tradition by manipulating their
learning experiences
(e.g. Gardner and
MacIntyre’s (1991) study
of instrumental
motivation).
designed to investigate
whether expected
patterns of relation-
ships involving IDs
and L2 learning occur
in specially selected
groups of learners
(e.g. Hansen and
Stansfield’s (1981) study
of the relationship
between different
cognitve styles and
L2 proficiency)

Figure 11.2: Types of ID research

Learners’ beliefs about language learning

Language learners—especially adults—bring a variety of beliefs to the class¬
room. According to Hosenfeld (1978), students form ‘mini theories’ of L2
learning. There has been relatively little research into the nature of these
theories and even less about how learners’ beliefs affect language learning.
Wenden (1986a; 1987a) reports a study of 25 adults enrolled in a part-time
advanced level class at an American university. She elicited their views about
language learning in a semi-structured interview and then summarized them
in terms of twelve explicit statements, grouped into three general categories.
The first category is ‘use of the language’. It includes beliefs relating to the im¬
portance of ‘learning in a natural way’—practising, trying to think in the L2,
and living and studying in an environment where the L2 is spoken. The se¬
cond category concerns beliefs relating to ‘learning about the language’.
Learners with beliefs in this category emphasized learning grammar and
vocabulary, enrolling in a language class, receiving feedback on errors
478 Explaining individual differences

they made, and being mentally active. The third category is labelled
‘importance of personal factors’. It includes beliefs about the feelings that
facilitate or inhibit learning, self-concept, and aptitude for learning. Wenden
found that her learners varied enormously in their beliefs, but that each
learner seemed to have a preferred set of beliefs that belonged to one of the
three categories.
Horwitz (1987a) used a questionnaire (The Beliefs about Tanguage
Learning Inventory) to elicit the beliefs of 32 intermediate level students of
different ethnic backgrounds who were studying on an intensive university
English program in the USA. Horwitz discusses the results in terms of five
general areas. Most learners (81 per cent) felt that people were born with a
special aptitude for learning foreign languages and were confident that they
possessed such an aptitude themselves! They believed that some languages
were more difficult than others, English being perceived as of average
difficulty. Many of the students held restricted views about the nature of
language learning, believing, for instance, that the best way to learn English
was to spend most of their time memorizing vocabulary and grammar rules.
With regard to culture, 94 per cent of the learners believed that it was
necessary to know something about English-speaking cultures in order to
speak English well. With regard to learning and communication strategies,
the learners favoured the use of audio materials and overwhelmingly
endorsed the need to repeat and practise. However, the learners varied in the
extent to which they felt it necessary to avoid errors. Finally, most of the
learners indicated that their main reason for learning English was to speak the
language fluently. They were keen on developing friendships with Americans,
but were also aware of the instrumental value of learning English.
Neither Wenden’s nor Horwitz’s study investigated the relationship
between learners’ beliefs and success in language learning. Little is currently
known about this. Abraham and Vann (1987) found some evidence that
beliefs might affect learning outcomes in a case study of two learners. Both
learners believed that it was important to create situations for using English
outside the classroom, to practise as much as possible, and to have errors
corrected. Both also believed it important to participate actively in class.
Gerardo, however, believed that paying conscious attention to grammar was
important, while Pedro did not and expressed a strong dislike of meta¬
language. Also, Gerardo thought that it was important to persevere in
communicating or understanding an idea, while Pedro considered topic
abandonment the best strategy in some cases. Abraham and Vann charac¬
terize Gerardo’s philosophy of language learning as ‘broad’ and Pedro’s as
‘narrow’. They suggest that this might have contributed to Gerardo’s better
TOEFL score (523 v. 473) at the end of a course of instruction. Pedro,
however, did better on a test of spoken English, which might suggest that
different views about language learning result in different kinds of success.
An interesting question is what determines learners’ beliefs about language
learning. Little, Singleton, and Silvius (1984, reported in Little and Singleton
 Individual learner differences 479

1990) surveyed random samples of undergraduate and postgraduate students

of foreign languages at Trinity College, Dublin. They found that ‘past experi¬
ence, both of education in general and of language learning in particular,
played a major role in shaping attitudes to language learning’ (1990: 14). For
example, the students stated that they preferred to learn by production activi¬
ties (repeating orally and writing) rather than through receptive activities
involving listening and reading. Little and Singleton claim that this belief re¬
flected the general nature of the instruction they had experienced. Learners’
beliefs are also likely to be influenced by general factors such as personality
and cognitive style, as Abraham and Vann’s (1987: 97) model of L2 learning
suggests. However, no studies have examined this relationship to date.
The study of learners’ opinions about language learning constitutes an im¬
portant area of enquiry, as it is reasonable to assume that their ‘philosophy’
dictates their approach to learning and choice of specific learning strategies.
Future research will need to find ways of ensuring that learners’ verbal reports
of their beliefs reflect their actual beliefs, and of investigating what effects dif¬
ferent ‘philosophies’ have on learning outcomes.

Learners’ affective states

Learners, in particular classroom learners, react to the learning situations
they find themselves in a variety of affective ways. For example, F. Schumann
(Schumann and Schumann 1977) reports being unable to settle down to stu¬
dying Farsi and Arabic (in Iran and Tunisia) until she had achieved order and
comfort in her physical surroundings. Bailey (1980) discusses a ‘classroom
crisis’ that occurred when her French teacher administered a test that the class
considered unfair. One of the beginner learners of German that Ellis and
Rathbone (1987) studied reported a period during which she was unable to
learn any German because of a boyfriend problem. These and other studies
testify to the complexity and dynamic nature of learners’ affective states and
the influence these have on their ability to concentrate on learning. Learners,
it seems, need to feel secure and to be free of stress before they can focus on the
learning task—the importance of which is directly acknowledged in human¬
istic approaches to language teaching (see Moskowitz 1978).
It is, of course, not easy to study the role that affective states play in L2
learning in a systematic manner, if only because they are so individualistic
and changeable. Some of the best work has investigated one particular af¬
fective factor—anxiety.

Anxiety
A distinction can be made between trait anxiety, state anxiety, and situation-
specific anxiety. Scovel (1978), drawing on work in general psychology, de¬
fines trait anxiety as ‘a more permanent predisposition to be anxious’. It is
480 Explaining individual differences

perhaps best viewed as an aspect of personality. State anxiety can be defined

as apprehension that is experienced at a particular moment in time as a re¬
sponse to a definite situation (Spielberger 1983). It is a combination of trait
and situation-specific anxiety. This latter type consists of the anxiety which is
aroused by a specific type of situation or event such as public speaking, exam¬
inations, or class participation.
The study of situational anxiety has received considerable attention in SLA
research (see, for example, Horwitz and Young’s (1991) collection of papers
on anxiety in classroom settings) and will be our main concern here. It has
been carried out concatenatively by means of correlational studies involving
measures of anxiety and learning (Type D research) and through diary studies
(Type B research). In both cases, the research indicates that learners fre¬
quently experience ‘language anxiety’, a type of situation-specific anxiety as¬
sociated with attempts to learn an L2 and communicate in it.
One of the key questions the research has addressed is ‘What causes state or
situation anxiety?’ The diary studies indicate that learners’ competitive na¬
tures can act as a source of anxiety. Bailey (1983) analysed the diaries of 11
learners and found that they tended to become anxious when they compared
themselves with other learners in the class and found themselves less profi¬
cient. She noted that as the learners perceived themselves becoming more pro¬
ficient, and therefore better able to compete, their anxiety decreased. Bailey
also identified other sources of anxiety, including tests and learners’ perceived
relationship with their teachers. Ellis and Rathbone (1987) reported that
some of their diarists found teachers’ questions threatening. For example, one
of their subjects, Monique, commented:

I was quite frightened when asked questions again. I don’t know why; the
teacher does not frighten me, but my mind is blocked when I’m asked ques¬
tions. I fear lest I give the wrong answer and will discourage the teacher as
well as be the laughing stock of the class maybe. Anyway, I felt really stupid
and helpless in class (Ellis 1989a: 257).

However, other learners experienced no worries about using German in class.

When anxiety does arise relating to the use of the L2, it seems to be restricted
mainly to speaking and listening, reflecting learners’ apprehension at having
to communicate spontaneously in the L2 (Horwitz, Horwitz, and Cope
1986). Matsumoto’s (1989) learner, M, for example, displayed little anxiety
but when she did it was in connection with her perceived problems in listening
to and speaking English. Learners can also experience anxiety as a result of
fear or experience of ‘losing oneself’ in the target culture. As Oxford (1992)
points out, this is closely related to the idea of ‘culture shock’. She lists the
affective states associated with this source of anxiety: ‘emotional regres¬
sion, panic, anger, self-pity, indecision, sadness, alienation, “reduced
personality”...’.
 Individual learner differences 481

These and other possible sources of anxiety in the foreign language class¬
room are reflected in the questionnaires which a number of researchers have
devised to measure learner anxiety. Gardner and Smythe (1975) developed a
French Class Anxiety Scale based on responses to a questionnaire, and in sub¬
sequent studies developed scales to tap English Use Anxiety and English Test
Anxiety. Horwitz, Horwitz, and Cope (1986) developed a Foreign Language
Classroom Anxiety Scale, based on conversations with beginner learners who
identified themselves as anxious. This questionnaire consists of thirty-three
items relating to three general sources of anxiety: (1) communication appre¬
hension, (2) tests, and (3) fear of negative evaluation. Examples of the kinds
of statements learners are asked to respond to on a five-point scale are:

I tremble when I know I am going to be called on in language class.

I keep thinking that other students are better at language than I am.
I get nervous when I don’t understand every word the language teacher
says.

MacIntyre and Gardner (1991b) have also developed an extensive set of

scales (23 in all) to measure various forms of anxiety.
The second main question which researchers have addressed is ‘What effect
does anxiety have on learning?’ MacIntyre and Gardner (1991b) point out
that anxiety can be hypothesized to affect three stages of the learning process:
input, processing, and output. However, most of the research to date has ex¬
amined output and so little is known about the effects on input and the pro¬
cess of learning. In an early study, Chastain (1975) found that measures of
anxiety in American university students were positively and significantly re¬
lated to marks in Spanish, positively but not significantly related to marks in
German, and negatively related to marks achieved by one group of audio-
lingual French learners, but not related at all to marks in another ‘regular’
French group. Kleinmann (1978) provides evidence to show that Spanish-
and Arabic-speaking learners of English who reported anxiety (i.e. responded
positively to statements like ‘Nervousness while using English helps me do
better’) were less likely to avoid complex grammatical structures such as in¬
finitive complements and the passive voice than those students who reported
no such anxiety.
Subsequent research has been based on models of language learning de¬
veloped by Gardner, Clement, and their colleagues (for example, Gardner,
Smythe, Clement, and Gliksman 1976; Clement and Kruidenier 1985).
MacIntyre and Gardner (1991b) found that what they term ‘language anxi¬
ety’ (the anxiety specifically generated by attempts to learn the L2) was negat¬
ively and significantly correlated with L2 performance on a Digit Span test
and on a French Categories test (which tested free recall of vocabulary items
associated with specific semantic fields), whereas no such correlation was
found between anxiety and these tests performed in the learners’ LI (English).
482 Explaining individual differences

MacIntyre and Gardner (1991a), in a comprehensive review of these studies,

conclude:

Covering several measures of proficiency, in several different samples, and

even in somewhat different conceptual frameworks, it has been shown that
anxiety negatively affects performance in the second language. In some
cases, anxiety provides some of the highest simple correlations of attitudes
with achievement (1991a: 103).

It should be noted, however, that not all the studies in their review produced
significant correlations between anxiety and achievement.
In fact, studies of learner anxiety have often produced even more mixed
results. Horwitz (1986) reports sizeable negative correlations (around -0.5)
between foreign language classroom anxiety and final grades achieved by
American university students. Young (1986) found that measures of anxiety
were correlated significantly with measures of oral proficiency in a group of
prospective language teachers, but the relationship disappeared after con¬
trolling for ability, measured by means of other proficiency measures.
Gardner, Moorcroft, and MacIntyre (1987) found a significant relationship
between various measures of anxiety and scores on a word production task,
but no relationship between the anxiety measures and free speech quality.
Parkinson and Howell-Richardson (1990) failed to find a relationship be¬
tween measures of anxiety, based on an analysis of fifty-one diaries kept by
adult learners of L2 English in Scotland, and rate of improvement—a result
they consider ‘surprising’. Ely (1986a) found no relationship between anxiety
and university learners’ level of participation in class, but the study did show
that those learners with a high level of discomfort were less likely to take risks
in class.
There are several reasons for these mixed results. One is that the relation¬
ship between anxiety and achievement is probably not a simple linear one.
Many of the above studies are premised on the assumption that learners with
low anxiety will learn better. Krashen (1981: 23) suggests how this might
come about:

The student who feels at ease in the classroom and likes the teacher may
seek out more intake by volunteering ... and may be more accepting of the
teacher as a source of input.

Scovel (1978), however, draws attention to Alpert and Elaber’s (1960) dis¬
tinction between facilitating and debilitating anxiety. The former motivates
learners to ‘fight’ the new learning task, prompting them to make extra efforts
to overcome their feelings of anxiety, although Horwitz (1986) suggests that
this may only occur in fairly simple learning tasks. The latter causes the learn¬
er to ‘flee’ the learning task in order to avoid the source of anxiety. Williams
(1991) suggests that the distinction between these two types of anxiety may
correspond to the intensity of the anxiety, with a low-anxiety state having a
 Individual learner differences 483

facilitating function and a high-anxiety state a debilitating effect. Also, the

two kinds of anxiety may sometimes cancel each other out, resulting in no ap¬
parent effect on achievement.
What is needed is a model to account for the role that anxiety plays in lan¬
guage learning, such as that proposed by MacIntyre and Gardner (1991a)
and summarized in Table 11.2. According to this model, the relationship be¬
tween anxiety and learning is moderated by the learners’ stage of develop¬
ment and by situation-specific learning experiences. The model also
recognizes that poor performance can be the cause as well as the result of
anxiety, a point also made by Skehan (1989).

Stage Type of anxiety Effect on learning

Beginner Very little—restricted to state anxiety None

Post-beginner Situation anxiety develops if learner Learner expects to be
develops negative expectations based on nervous and performs poorly
bad learning experiences
Later Poor performance and continued bad Continued poor performance
learning experiences result in increased
anxiety

Table 11.2: A model of the role of anxiety in language learning (based on

MacIntyre and Gardner 1989a)

There is sufficient evidence to show that anxiety is an important factor in

L2 acquisition. Anxiety (its presence or absence) is best seen not as a neces¬
sary condition of successful L2 learning, but rather as a factor that contrib¬
utes in differing degrees in different learners. The work done to date has gone
only a limited way to determine the conditions under which anxiety will have
an effect. It has also relied exclusively on self-report questionnaires rather
than on the measurement of somatic responses (for example, galvanic skin
response).1

The importance of learners’ affective states

Learners’ affective states are obviously of crucial importance in accounting
for individual differences in learning outcomes. Whereas learners’ beliefs
about language learning are likely to be fairly stable, their affective states tend
to be volatile, affecting not only overall progress but responses to particular
learning activities on a day-by-day and even moment-by-moment basis.
Studies in the naturalistic research tradition may prove most effective in
exploring how these transitional states are brought about and what effect
they have on learning.
484 Explaining individual differences

General factors contributing to individual differences in second

language learning
Learners’ beliefs and affective states are likely to have a direct effect on L2
learning, but they themselves may be influenced by a number of general fac¬
tors relating to learners’ ability and desire to learn and the way they choose to
go about learning. In this section we will examine in some detail the general
factors that have received the most attention in SLA research.

Age

There is a widely-held lay belief that younger L2 learners generally do better

than older learners. This is supported by the critical period hypothesis, ac¬
cording to which there is a fixed span of years during which language learning
can take place naturally and effortlessly, and after which it is not possible to
be completely successful. Penfield and Roberts (1959), for example, argued
that the optimum period for language acquisition falls within the first ten
years of life, when the brain retains its plasticity. Initially, this period was
equated with the period taken for lateralization of the language function to
the left side of the brain to be completed. Work on children and adults who
had experienced brain injuries or operations indicated that damage to the left
hemisphere caused few speech disorders and was rapidly repaired in the case
of children but not adults (Lenneberg 1967). Although subsequent work (for
example, Krashen 1973; Whitaker, Bub, and Leventer 1981) has challenged
the precise age when lateralization takes place, resulting in doubts about the
neurological basis of the critical period hypothesis, the age question has con¬
tinued to attract the attention of researchers.
It continues to be controversial, however (see, for example, the papers by
Flege (1987) and Patkowski (1990) in Applied Linguistics). The controversy
centres on both whether there are significant differences in L2 learning ac¬
cording to age, and also on the theoretical explanations for those differences
which researchers claim to have found.
One of the reasons for the lack of consensus on the age issue is undoubtedly
the difficulty of comparing the results of studies that have employed very dif¬
ferent methods. There are longitudinal studies based on groups of learners
with the same starting time (for example, Snow and Hoefnagel-Hohle 1978)
and also longitudinal studies based on groups of learners with different start¬
ing ages (for example, Burstall 1975). There are cross-sectional studies of
groups of learners who differ with regard to both when individual learners
began to learn the L2 and, inevitably, the number of years they have been
learning it (for example, Oyama 1976). Finally, there are experimental stud¬
ies, which have sought to investigate the effects of attempts to teach groups of
learners varying in age-specific features of an L2 (for example, Neufeld
1978). The studies have also varied in how they have measured learning. In
 Individual learner differences 485

some cases, performance measures based on samples of planned or

unplanned language use have been used. In others, the grammaticality judge¬
ments of learners of different ages have been examined (for example,
Coppieters 1987), while in still others native speakers have been asked to rate
the performance of mixed groups of learners and native speakers in terms of
how ‘native’ their use of the language is (for example, Scovel 1981). It is per¬
haps not surprising that the results obtained by these studies fail to agree.
As Larsen-Freeman and Long (1991) point out, however, the age issue re¬
mains an important one for theory building in SLA research, for educational
policy-making, and for language pedagogy. If it can be shown that older
learners are different from younger learners, the claim that adults have con¬
tinued access to Universal Grammar is called into question. If it can be shown
that younger learners do better than older learners, the case for an early start
in foreign language education is strengthened. If it can be shown that children
learn in different ways to adults, language teachers will need to identify differ¬
ent approaches and techniques to suit the two kinds of learners.
In order to untangle the research results, it is helpful to consider a number
of separate but related questions:

1 What effect does age have on the rate of L2 learning?

2 What effect does age have on learners’ ability to achieve native-speaker
levels of proficiency?
3 What effect does age have on learners’ levels of L2 achievement (in those
learners who do not reach native-speaker proficiency)?
4 What affect does age have on the processes of L2 learning?

It is also helpful to distinguish the effects of age according to learning con¬

text (in naturalistic as opposed to instructed situations). Finally, the effects of
age on the learning of L2 pronunciation and L2 grammar need to be con¬
sidered separately. Unfortunately, there has been very little research that has
investigated other aspects of L2 learning.

The effects of age on rate of second language learning

In their review of the research that has addressed the age issue, Krashen,
Long, and Scarcella (1979) conclude that (1) adults are superior to children in
rate of acquisition, and (2) older children learn more rapidly than younger
children. The study most often cited in support of these conclusions is Snow
and Hoefnagel-Hohle (1978). This study investigated the naturalistic acquisi¬
tion of Dutch by eight- to ten-year-old English-speaking children, twelve- to
fifteen-year-old adolescents, and adults over a ten-month period. The learn¬
ers’ proficiency was measured on three separate occasions (after three
months, six months, and at the end of the study). With regard to morphology
and syntax the adolescents did best, followed by the adults, with the children
last. However, there were only small differences in pronunciation, and the
486 Explaining individual differences

grammar differences diminished over time as the children began to catch up.
Experimental studies have also shown that adults outperform children in the
short term. For example, Olsen and Samuels (1973) found that American
English-speaking adolescents and adults performed significantly better than
children after ten 15-25 minute German pronunciation sessions. However,
other studies suggest that, at least where pronunciation is concerned, adults
do not always progress more rapidly than children. Cochrane (1980), for ex¬
ample, investigated the ability of 54 Japanese children and 24 adults to dis¬
criminate English /r/ and /!/. The average length of naturalistic exposure was
calculated as 245 hours for the adults and 193 for the children (i.e. relatively
little). The children outperformed the adults, although in a follow-up experi¬
ment in which the two groups were taught the phonemic distinction, the ad¬
ults benefited while the children did not. The research gives general support to
Krashen, Long and Scarcella’s generalization that adults learn faster than
children. It appears to be more applicable to grammar than pronunciation
(where children seem to learn as rapidly, if not more rapidly, than adults), al¬
though in the case of formal learning situations adults seem to do better even
in this area of learning. It is not yet clear at what point children start to catch
up.

The effects of age on the acquisition of native-speaker proficiency

The controversy regarding the role of age is fiercest when it comes to con¬
sidering the effects of age on the achievement of native-speaker levels of profi¬
ciency. This question is the crucial one for the critical period hypothesis.
Neufeld’s (1978) study is often cited by those seeking evidence to refute the
hypothesis. In this study, 20 adult native speakers of English were given 18
hours of intensive instruction in the pronunciation of Chinese and Japanese.
To test the ‘nativeness’ of their pronunciation, the learners were then given an
imitation test and their utterances judged on a five-point scale (from ‘unmis¬
takably native’ to ‘heavily accented’) by native speakers of the two languages.
Nine and eight of the subjects were rated as ‘native’ for Japanese and Chinese
respectively. This study suggests, therefore, that under the right conditions
adults can achieve native ability in pronunciation—the area of language gen¬
erally considered to be the most difficult for adults to acquire. Neufeld (1977;
1979) conducted other studies with similar results. However, his studies have
been strongly criticized by supporters of the critical period hypothesis. Long
(1990a), for instance, argues that Neufeld’s subjects represented an ‘elite’,
that the imitation test produced ‘rehearsed’ rather than natural data, and that
the instructions given to the raters predisposed them to think that some of the
subjects were native speakers. These criticisms—and those made by Patkow-
ski (1990)—are legitimate, but they do not refute the essential claim that
Neufeld seeks to make—namely that it is possible for adults to achieve
 Individual learner differences 487

native-speaker levels of proficiency in an L2. Clearly the study needs further

replication.
Another frequently cited experimental study provides evidence to support
the critical period hypothesis. Coppieters (1987) tested 21 highly proficient
speakers of French, all of whom had begun learning as adults, and compared
their performance on a grammaticality judgement task with that of 20 native
speakers. Coppieters notes that it was not possible to distinguish the two
groups by the mistakes they made, their choice of lexis, or grammatical con¬
structions, and six of the subjects were also described as having no traces of a
foreign accent. The results of the grammaticality judgement test, however,
showed clear differences between the two groups, suggesting that despite the
native-like performance of the learners in language production, their gram¬
matical competence differed from that of native speakers.2 Again, though, it
is possible to raise methodological objections to this study. Coppetiers did
not include a group of learners who had started to learn L2 French as chil¬
dren, thus we cannot be sure that the results he obtained reflect age as op¬
posed to some other factor. Also, as in the case of Neufeld’s imitation test,
doubts can be raised about whether grammaticality judgements constitute a
valid means of measuring competence (see the discussion of grammaticality
judgement tasks in Chapter 10). Birdsong (1992) identifies ‘numerous pro¬
cedural and methodological features of the Coppetiers study that comprom¬
ise its conclusions’ (1992: 711).
Birdsong’s own replication of this study casts serious doubts on the results
Coppetiers obtained. Birdsong administered a grammaticality judgement test
to 20 English-speaking learners of L2 French, who were near-native in their
oral ability, and to 20 native speakers of French. The study was motivated by
Long’s (1990a) challenge to researchers to investigate ‘whether the very best
learners actually have native-like competence’ (1990a: 281). Contrary to
Coppieters, Birdsong found no evidence of any dramatic differences in the
judgements of the non-native speakers and native speakers. A number of the
non-native spearkers performed in the same range as the native speakers on
the grammaticality judgement test. Furthermore, Birdsong could find no
evidence of marked differences between the two groups in the think-aloud
data that he collected from the subjects as they performed their judgements.
This study, then, suggests that at least some learners who start learning a L2
after puberty achieve a level of competence indistinguishable from that of
native speakers.
Another way of investigating the claims of the Critical Period Hypothesis is
to investigate whether learners who start learning an L2 as young children
and enjoy favourable learning conditions succeed in reaching native levels of
proficiency. Thompson’s (1991) study of foreign accents in Russian immig¬
rants in the United States addressed this question.3 Thompson found that
those learners who had arrived before they were ten years old had a more
native-like English accent than those who came after this age—a finding that
488 Explaining individual differences

bears out the results of earlier studies reported in the next section. What is in¬
teresting about this study, though, is that two subjects who came to the US at
the age of four years were still rated as having a ‘slight accent’, a result that
Thompson considers ‘a problem for the Critical Period Hypothesis’ (1991:
199). Thompson speculates that these learners’ failure to achieve native-
speaker levels of pronunciation was because they had maintained a high level
of speaking proficiency in Russian, and that this led to what Weinreich (1953)
has called an ‘interlingual’ identification. Thompson’s study is important be¬
cause it suggests the need to consider age in relation to other factors, such as
LI maintenance, and that not all learners will wish to sound like native
speakers.
Yet another way of assessing whether learners can achieve native-speaker
levels in an L2 is to see whether they are able to recognize spoken or written
‘accents’ in the same way as native speakers. Scovel (1981) asked four groups
of judges (adult native speakers, child native speakers, adult non-native
speakers, and adult aphasics) to rate speech samples and written pieces pro¬
duced by a mixture of native and non-native speakers. He found that even the
most advanced non-native speakers achieved an accuracy rate of only 77 per
cent, which was about the same as the child native speakers (73 per cent) but
less than the adult native speakers (95 per cent) and even the aphasic native
speakers (85 per cent). Like Coppieters’ study, this study suggests that even
very advanced learners lack some of the linguistic abilities of native speakers.
The experimental studies that have investigated the effects of age on the ac¬
quisition of native-speaker levels of proficiency have produced mixed results
and, at this stage, the verdict must remain an open one. It is possible that un¬
der ideal circumstances learners who start after puberty can learn to produce
speech and writing that cannot easily be distinguished from that of native
speakers. Whether qualitative differences in competence still remain, as
claimed by Coppieters, is still not clear, although Birdsong’s carefully de¬
signed study would suggest that at least some learners achieve native-speaker
levels of grammatical knowledge. Also, as Thompson’s (1991) study shows,
starting early is no guarantee that native-speaker abilities will be achieved,
even in the most favourable learning situations.

The effects of age on learners’ second language achievement

The majority of L2 learners fail to reach native-speaker levels of ability. It is
also important to ask whether age effects are evident in such learners. Do
learners who begin learning as children in general reach higher levels of L2
ability than those who start as adolescents or adults? This question has been
addressed in research that has compared the level of proficiency reached by
L2 learners who began as children with that of learners who began as adults.
We do not know, of course, if these studies show the effects of age on these
 Individual learner differences 489

learners’ ultimate level of attainment, as the assumption that they have

reached their ‘final state’ (are fossilized) may not be justified.
A number of studies have investigated the relative effects of starting foreign
language education in the primary school as opposed to the secondary school
on the levels of attainment. For example, Burstall (1975) reports on a pilot
scheme in England and Wales. She compared two groups of students with five
years of instruction. One group had begun learning French at the age of 8,
while the other had begun at the beginning of secondary school (11 years).
She found that the older learners were ‘consistently superior’. When both
groups were compared at the age of 16, the secondary school starters out¬
performed the primary school starters on tests of speaking, reading, and writ¬
ing and were inferior only on a test of listening. Harley (1986) investigated the
levels of attainment of children in French bilingual programmes in Canada.
She focused on the learners’ acquisition of the French verb system, obtaining
data from interviews, a story repetition task, and a translation task. She com¬
pared early and late immersion students after both had received 1,000 hours
of instruction. Neither group had acquired full control of the verb system, but
the older students demonstrated greater overall control. However, the early
immersion group showed higher levels of attainment at the end of their
schooling, a result that may reflect the additional number of years’ instruction
they had received rather than starting age. The results from these and other
school-based studies (see Singleton (1989) for a review) is not supportive of
the claim that children’s level of attainment is greater than that of adolescents/
adults. One possible explanation for this—advanced by Singleton—is that
formal learning environments do not provide learners with the amount of ex¬
posure needed for the age advantage of young learners to emerge.
Studies of learners in naturalistic learning situations provide the most con¬
vincing evidence that younger is better and, therefore, some support for the
Critical Period Hypothesis. We will examine some of the most frequently
cited studies.
Fearners who start as children achieve a more native-like accent than those
who start as adolescents or adults. Oyama (1976) investigated 60 male im¬
migrants who had entered the United States at ages ranging from 6 to 20 years
and had been resident there for between 5 and 18 years. She asked two adult
native speakers to judge the nativeness of the learners’ accents in two 45-
second extracts taken from performance on a reading-aloud task and a free-
speech task. Oyama reports a very strong effect for ‘age of arrival’ but almost
no effect for ‘number of years’ in the United States. She found that the young¬
est arrivals performed in the same range as native-speaker controls. Other
studies which have investigated the effects of age on pronunciation (for ex¬
ample, Asher and Garcia 1969; Tahta, Wood, and Foewenthal 1981) support
the younger-is-better position.
Similar results have been obtained for the acquisition of grammar. Patkow-
ski’s (1980; 1990) study of 67 educated immigrants to the United States
490 Explaining individual differences

found that learners who had entered the United States before the age of 15
were rated as more syntactically proficient than learners who had entered af¬
ter 15. Furthermore, there was a marked difference in the distribution of the
scores (based on native speakers’ ratings on a five-point scale) for the two
groups. The adult group’s scores were evenly distributed, with the majority at
midpoints on the rating scale. The child group’s scores clustered at the high
end of the rating scale, with 29 out of 33 achieving a rating of 4+ or 5. Pat-
kowski also investigated the effects of number of years spent in the United
States, amount of informal exposure to English, and amount of formal in¬
struction. Only the amount of informal exposure had any significant effect,
and even this was negligible in comparison with the age factor. Patowski’s
findings are confirmed by Johnson and Newport’s (1989) study of 46 native
Koreans and Chinese who had arrived in the United States between the ages
of 3 and 39, half before the age of 15 and half after 17. The subjects were
asked to judge the grammaticality of 276 spoken sentences, about half of
which were grammatical. Overall the correlation between age at arrival and
judgement scores was -0.77 (i.e. the older the learners were at arrival, the
lower their scores). Far less variation was found in the scores of the ‘child’
group than in the adult group. Neither the number of years of exposure to
English beyond five nor the amount of classroom instruction was related to
the grammaticality judgement scores, and although an effect for ‘identifica¬
tion with American culture’ was found, this was much weaker than that for
age.
In his summary of these and other studies, Singleton (1989) writes:
Concerning the hypothesis that those who begin learning a second lan¬
guage in childhood in the long run generally achieve higher levels of profi¬
ciency than those who begin in later life, one can say that there is some good
supportive evidence and that there is no actual counter evidence (1989:
137).
This is one of the few definite conclusions that Singleton feels able to reach in
a comprehensive survey of age-related research. It is worthwhile noting, how¬
ever, that this conclusion may not hold true for the acquisition of L2 literacy
skills. Cummins and Nakajima (1987) examined the acquisition of reading
and writing skills by 273 Japanese children in grades two to eight in Toronto.
They found that the older the students were on arrival in Canada, the more
likely they were to have strong L2 reading skills and, to a lesser extent, better
L2 writing skills. The explanation Cummins and Nakajima offer is that the
older learners benefited from prior literacy experience in Japanese (see the
discussion of the Interdependency Principle in Chapter 6).

The effect of age on the process of second language acquisition

There have been few studies of the effects of age on the process of L2 acquisi¬
tion. The morpheme studies (see Chapter 3) showed that the order of
 Individual learner differences 491

acquisition of a group of English morphemes was the same for children and
adults (Bailey, Madden, and Krashen 1974; Fathman 1975). However,
conclusions based on the morpheme studies are circumspect given their
methodological problems. Studies which have investigated the sequence of
acquisition in transitional structures such as negatives and interrogatives are
not subject to the same methodological strictures, however. They show that
adults go through the same stages of acquisition as children (for example,
Cancino et al. 1978). Age, therefore, does not appear to affect the general
developmental pattern.
By far the most detailed study of the effects of age on the acquisition pro¬
cess is Harley’s (1986) investigation of early and late immersion programmes.
Harley found remarkably similar patterns in the two groups’ acquisition of
the French verb phrase. For example, the two age groups generally made sim¬
ilar types of errors and both groups tended to use the relatively unmarked
French verb forms more accurately than the marked forms. A few differences
were noted but these were minor, and Harley did not feel that they consti¬
tuted evidence of different mental processes, arguing instead that the differ¬
ences reflected variations in the F2 input to which the learners were exposed.
Process differences may occur in F2 pronunciation, however. Riney (1990)
reviewed literature relating to whether learners display a preference for an
open syllable structure in early interlanguage. He argued that in the case of
learners who began before the age of 12 years, no open syllable preference is
evident (as Sato’s (1987) study indicates), but in the case of learners beginning
after 12 years there was, as in Tarone’s (1980a) study. In data collected from
Vietnamese learners of English, Riney was able to show that whereas age had
no effect on the final deletion of consonants (one way of making a target-
language closed syllable open), it did have a marked effect on epenthesis (the
insertion of a vowel at the end of a closed syllable). Whereas the incidence of
epenthesis in 10-12-year-old children was less than 5 per cent, in some adult
learners it was over 30 per cent. Furthermore, epenthesis in adult learners did
not significantly decline with increased exposure to English.
It is obviously premature to conclude that age has no effect on the process
of acquisition. The research to date suggests that the effect may be a minimal
one in the case of grammar, but possibly more significant in the case of
pronunciation.

Some general conclusions

The research that has addressed the age issue is quite enormous. Not surpris¬
ingly, commentators have arrived at different conclusions, but despite this
some common ground is emerging:
1 Adult learners have an initial advantage where rate of learning is con¬
cerned, particularly in grammar. They will eventually be overtaken by
child learners who receive enough exposure to the F2. This is less likely
492 Explaining individual differences

to happen in instructional than in naturalistic settings because the critical

amount of exposure is usually not available in the former.
2 Only child learners are capable of acquiring a native accent in informal
learning contexts. Long (1990a) puts the critical age at 6 years, but Scovel
argues that there is no evidence to support this and argues for a pre-puberty
start. Singleton (1989) points out that children will only acquire a native
accent if they receive massive exposure to the L2. However, some children
who receive this exposure still do not achieve a native-like accent, possibly
because they strive to maintain active use of their LI. Adult learners may be
able to acquire a native accent with the assistance of instruction, but
further research is needed to substantiate this claim.
3 Children may be more likely to acquire a native grammatical competence.
The critical period for grammar may be later than for pronunciation
(around 15 years). Some adult learners, however, may succeed in acquiring
native levels of grammatical accuracy in speech and writing and even full
‘linguistic competence’.
4 Irrespective of whether native-speaker proficiency is achieved, children are
more likely to reach higher levels of attainment in both pronunciation and
grammar than adults.
5 The process of acquiring an L2 grammar is not substantially affected by
age, but that of acquiring pronunciation may be.

Explaining the role of age in second language acquisition

These general conclusions provide substantial support for the existence of at
least a sensitive period for L2 acquisition. The distinction between a ‘critical’
and a ‘sensitive’ period rests on whether completely successful acquisition is
deemed to be only possible within a given span of a learner’s life, or whether
acquisition is just easier within this period. The conclusions also lend some
support to Seliger’s (1978) proposal that there may be multiple critical/
sensitive periods for different aspects of language. The period during which a
native accent is easily acquirable appears to end sooner than the period gov¬
erning the acquisition of a native grammar.
A number of explanations have been advanced to account for the existence
of a critical or sensitive period. These have been admirably reviewed in
Singleton (1989) and Long (1990a), and are summarized in Table 11.3.
Singleton points out the problems that exist with all the explanations and
declines to come down in favour of one. One of the major points of contro¬
versy is whether the differences between child and adult learners are to be
explained as primarily the result of environmental factors or of changes in the
mental and neurological mechanisms responsible for language learning.
Miihlhauser (1986), after an extensive study of the developmental stages of
pidgin languages and their similarities to language acquisition, concludes that
‘adults and children appear to behave very much in the same manner’, which
 Individual learner differences 493

indicates that activation of certain linguistic developments is dependent on

the presence of specific environmental factors, rather than on different cog¬
nitive abilities of children and adults’ (1986: 265-6). Long, on the other
hand, concludes that a neurological explanation is best and proposes the
attractive-sounding ‘mental muscle model’, according to which ‘the lan¬
guage-specific endowment remains intact throughout adult life, but access to
it is impeded to varying degrees and progressively with age, unless the faculty
is used and so kept plastic’. Such a view is compatible with studies of excep¬
tional language learners (for example, Obler 1989), which demonstrate that
some adult learners are capable of achieving native-speaker levels of compet¬
ence. As Birdsong (1992) points out, the question then arises as to whether it
is possible to maintain the Critical Period Hypothesis if many such learners
are found.
One puzzle is why there is so little evidence of any differences in the process
of L2 acquisition by child and adult learners. If adults substitute inductive
cognitive learning strategies for the language acquisition device used by chil¬
dren, differences in the process of acquisition might be expected to occur. In
the case of phonology, some clear process differences have been reported,
suggesting that children and adults rely on different mechanisms. However,
in the case of grammar no clear differences have been observed, suggesting
that learners of all ages rely on the same learning mechanisms. Long’s ‘mental
muscle model’, therefore, may not provide a satisfactory explanation where
L2 phonology is concerned, but seems to offer a convincing account of why
child and adult learners do not differ in the process of acquiring an L2
grammar.
It is not necessary to posit neurological explanations to account for why
older learners learn more rapidly. One possibility, which we considered in
Chapter 6, is that older learners experience more negotiation of meaning and,
therefore, better input. Another obvious possibility is that adolescents and
adults possess more fully developed cognitive skills, which enable them to ap¬
ply themselves studially to the task of learning a L2. This is likely to give them
an initial advantage over children, but may not be sufficient to guarantee high
levels of L2 proficiency. Most likely, the rate advantage enjoyed by adults is
the result of a combination of factors.
To conclude, it is not yet possible to reach any definite decisions on such
key issues as whether adults have continued access to a language-specific
acquisition device such as Universal Grammar. One tentative conclusion
suggested by the research is that the acquisition of phonology (which appears
to be particularly sensitive to age) proceeds somewhat differently from the
acquisition of grammar (which appears much less sensitive). This conclusion,
it should be noted, accords with the conclusion we reached when considering
the role of LI transfer in L2 learning (see Chapter 8).
494 Explaining individual differences

Type of explanation Main arguments

1 Sensory acuity The language learning capacity of adults is impaired by

deterioration in their ability to perceive and segment sounds in
an L2.
2 Neurological There are changes in the neurological structure of the brain at
certain ages which affects learners’ abilities to acquire L2
pronunciation and grammar. Various accounts of the nature of
these changes have been proposed to account for the ‘loss of
plasticity’ that occurs with age (e.g. lateralization and cerebral
maturation).
3 Affective- Child learners are more strongly motivated to communicate with
motivational factors native speakers and to integrate culturally. Also, child learners
are less conscious and therefore suffer less from anxiety about
communicating in an L2.
4 Cognitive factors Adult learners rely on general, inductive learning abilities to learn
an L2, while children use their language acquisition device.
5 Input The language input received by children is superior to that
received by adults. However, adults may experience more
negotiation of meaning.
6 Storage of L2 Children store LI and L2 information separately (i.e. become
information coordinate bilinguals); adults store LI and L2 knowledge
together (i.e. become compound bilinguals).

Table 11.3: Explanations of the role of age in L2 acquisition

Language aptitude

In an article reviewing early aptitude research, Carroll (1981) defines general

aptitude as ‘capability of learning a task’, which depends on ‘some combina¬
tion of more or less enduring characteristics of the learner’. In the case of lan¬
guage aptitude the capability involves a special propensity for learning an L2.
The general claim that language aptitude constitutes a relevant factor in L2
acquisition entails, in Carrol’s view, a number of more specific claims. The
first is that aptitude is separate from achievement. Carroll argues that they are
conceptually distinct and also that they can be distinguished empirically (by
demonstrating that there is no relationship between measures of aptitude and
measures of proficiency at the beginning of a language programme, but that
there is a relationship at the end of the programme). Second, aptitude must be
shown to be separate from motivation. On this point, however, there is some
disagreement, as Pimsleur (1966) treats motivation as an integral part of apti¬
tude. Carroll argues that research by Lambert and Gardner (reviewed later
in this chapter) has consistently shown that aptitude and motivation are
 Individual learner differences 495

separate factors. Third, aptitude must be seen as a stable factor, perhaps even
innate. In support of this claim, Carroll refers to studies which show that
learners’ aptitude is difficult to alter through training. Fourth, aptitude is to
be viewed not as a prerequisite for L2 acquisition (as all learners, irrespective
of their aptitude, may achieve a reasonable level of proficiency), but as a
capacity that enhances the rate and ease of learning. Aptitude tests, therefore,
provide a prediction of rate of learning.
Finally, Carroll argues that aptitude must be found to be distinct from gen¬
eral intelligence. He refers again to research by Lambert and Gardner which
has shown that aptitude and intelligence measurements are not related. There
are doubts about this claim, however. Pimsleur considers intelligence an im¬
portant part of aptitude. Oiler and Perkins (1978) have also argued that ver¬
bal intelligence is a major factor as it is needed to answer tests of the kind used
to measure aptitude and language proficiency and thus is a common factor to
both. In contrast, although finding significant correlations between scores on
a verbal intelligence test and a test of foreign language proficiency, Skehan
(1990) argues that there are clear differences between them. Similarly, Obler
(1989), in a study of one exceptional learner who had a record of‘picking up’
languages with great rapidity and ease, concluded that ‘generally superior
cognitive functioning is not necessary for exceptional L2 acquisition’ (1989:
153). We will return to the question of intelligence later.
The two main instruments used to measure aptitude were developed in the
1950s and 1960s. The Modern Language Aptitude Test (MLAT) (Carroll
and Sapon 1959) was developed initially as a means of screening candidates
for foreign language instruction at the Foreign Service Institute in the United
States. It exists in various forms. There is a full form, a short form, and an
MLAT-Elementary, which was designed for selecting, guiding, and placing
children in the Foreign Language in the Elementary School (FLES) pro¬
gramme. Versions have also been developed for use with languages other
than English—French, Italian, and Japanese. The Pimsleur Language Apti¬
tude Battery (PLAB) (Pimsleur 1966) was developed as an alternative to
MLAT and, in particular, with a view for use in the junior high school. It
measures a very similar range of abilities to MLAT. Other less well-known
aptitude tests include the Defence Language Aptitude Battery (Petersen and
Al-Haik 1976). This tested learners’ ability to learn an artificial language
through auditory and visual materials. It was designed for use with learners at
the higher ends of the ability range. The York Language Aptitude Test (Green
1975) tested the ability of learners to use analogy to produce forms in an un¬
known language (Swedish). MLAT and PLAB are the most commonly used in
aptitude research.
Carroll (1965) identified four factors in language aptitude:
496 Explaining individual differences

1 Phonemic coding ability (the ability to code foreign sounds in a way that
they can be remembered later). This ability is seen as related to the ability to
spell and to handle sound-symbol relationships.
2 Grammatical sensitivity (the ability to recognize the grammatical functions
of words in sentences).
3 Inductive language learning ability (the ability to identify patterns of
correspondence and relationships involving form and meaning).
4 Rote learning ability (the ability to form and remember associations be¬
tween stimuli). This ability is hypothesized to be involved in vocabulary
learning.

Somewhat surprisingly, however, Carroll and Sapon’s MLAT does not in¬
clude a separate measure of (3), inductive language learning ability, perhaps
because this is very close to (2), grammatical sensitivity. The PLAB covers a
similar range of characteristics, but has no test of verbal memory. Also, in this
test a score is awarded for ‘Grade point average in academic areas’, as a meas¬
ure of intelligence.
Aptitude research has adopted a concatenative approach. Measures of lan¬
guage aptitude are correlated with measures of language proficiency and
achievement. The main aim of the research has been to establish to what ex¬
tent it is possible to predict learning outcomes. A secondary aim has been to
identify which specific aspects of aptitude individual learners are strongest in,
so that matching forms of instruction can be provided ('Wesche 1981). A gen¬
eral assumption of the research is that aptitude will only have an effect on
learning outcomes if the learners are sufficiently motivated to learn (i.e. make
the effort to use their intrinsic abilities).
A fairly strong relationship has been consistently found between language
aptitude and learning. Carroll (1981) reports that the studies he carried out
using the MLAT produced correlations between .40 and .60 with a variety of
criterion measures (final course grades, objective foreign language attainment
tests, and instructors’ estimates of learners’ language learning abilities).
Gardner (1980), in a review of several studies of the effects of motivation and
aptitude on the learning of French in schools throughout Canada, reports a
median correlation for aptitude of .41. Pimsleur, Sundland, and MacIntyre
(1966) emphasize the importance of sound discrimination, having found that
20 to 30 per cent of the children in their study were underachieving because of
particular problems with this component of aptitude. With the exception of a
study by Horwitz (1987b), and Skehan’s recent work on aptitude, however,
there has been little research since the 1970s.
Horwitz’s study is interesting because it suggests that language aptitude is
related to measures of both linguistic and communicative competence. It in¬
vestigated 61 female high school students of French in the United States. Apti¬
tude scores on MLAT correlated significantly with scores on a discrete-point
 Individual learner differences 497

written grammar test (0.41) and with scores on a series of oral tasks requiring
relatively spontaneous language (0.40).
Skehan’s research (1986a, 1986b, and 1990) has been directed at two gen¬
eral issues. One concerns the nature of the abilities which aptitude tests meas¬
ure. The other is whether aptitude is best seen as a cumulative aggregation of
abilities or as differentiated, thus affording more than one route to success.
In order to re-examine the nature of aptitude, Skehan used data from
Wells’ (1985) longitudinal study of the LI acquisition of English by some
sixty plus children in the Bristol area of England. He investigated (1) the rela¬
tionship between measures of LI development and standard measures of ap¬
titude, (2) the relationship between measures of foreign language learning by
the same children at 13 years of age and aptitude, and (3) the relationship be¬
tween LI development and foreign language achievement. A number of signi¬
ficant relationships involving measures of LI development and aptitude were
found. The highest correlations were between measures of fragile syntax (for
example, auxiliary and pronoun systems) and analytic aspects of aptitude.
Language aptitude was also strongly related to foreign language achieve¬
ment. However, only test-based and lexical measures of LI development
were related to the measures of foreign language achievement, which were de¬
rived exclusively from tests. Skehan explains these results by arguing that the
aptitude tests he used measured two main aspects: (1) an underlying language
learning capacity, which is similar in LI and L2 learning, and (2) an ability to
handle decontextualized material, such as that found in formal language
tests. This research constitutes an important extension of the earlier research
because it shows that aptitude involves both the kind of abilities Carroll and
Pimsleur had identified earlier and, additionally, the ability to deal with con¬
text-free language.
Skehan’s research also helps to illuminate the question of whether intelli¬
gence is to be seen as part of aptitude (as Pimsleur and Oiler claimed) or as dis¬
tinct from it (as Carroll claimed). Here Cummins’ (1983) distinction between
basic interpersonal communication skills (BICS) and cognitive academic lan¬
guage proficiency (CALP) is relevant (see Chapter 6). The former consists of
those skills required for oral fluency and the sociolinguistically appropriate
use of a language. The latter concerns a dimension of language proficiency
that is strongly related to overall cognitive and academic language skills and
can be equated with the global language proficiency factor, which Oiler and
Perkins (1978: 413) have claimed accounts for ‘the bulk of the reliable vari¬
ance in a wide range of language proficiency measures’ and is identical with
the ‘g’ factor of intelligence. A number of studies support a connection be¬
tween CALP and intelligence. Genesee (1976) found that intelligence was
strongly related to the development of academic L2 French language skills
(reading, grammar, and vocabulary) but was largely unrelated to ratings of
oral productive ability. Ekstrand (1977) also found low-level correlations
498 Explaining individual differences

between intelligence and proficiency as measured on tests of listening com¬

prehension and free oral production, but much higher correlations when
proficiency was measured by tests of reading comprehension, dictation, and
free writing. The underlying learning capacity which Skehan found can be hy¬
pothesized to relate to BICS, a claim that is supported by the strong relation¬
ship between aptitude and LI acquisition (which takes place in the course of
developing BICS). The ability to handle decontextualized language, however,
suggests that aptitude is also related to CALP, and therefore also involves in¬
telligence. So, in a sense, Carroll and Pimsleur were both right.
Skehan’s research has also addressed another issue—the extent to which
the effect of aptitude is to be viewed ‘globally’, as the aggregation of aptitude
strengths in the different components, or differentially, with learners finding
different routes to success in language learning depending on the nature of
their aptitude. Skehan (1986a) reports a study in which he used cluster ana¬
lysis to identify different kinds of learners who were studying colloquial Ar¬
abic in the Army School of Languages in Britain. The results were not entirely
clear, but they suggested that whereas some learners were grammatically
sensitive and demonstrated finely-tuned inductive language learning ability,
others were strong on memory and ‘chunk-learning’. This led Skehan (1989)
to propose that there are two types of foreign language learners, analytic and
memory-orientated. Both kinds can achieve a high level of success. Studying
the differences in the type of aptitude that individual learners have serves as
one way of investigating learning style (see the next section).
Research into language aptitude was popular in the 1960s and 1970s, but
has largely gone into abeyance since. One reason for this is undoubtedly the
strong link that existed between the models of aptitude developed during that
time and the prevailing views of language (structuralist), language learning
(behaviourist), and language teaching (audiolingual). As these views were
challenged, so interest in aptitude declined. The research was restricted in
that it only looked at learners in formal learning situations, and in general
tended to use criterion measures of learning that reflected CALP rather than
BICS. Despite these limitations, the early research provided convincing evid¬
ence that classroom learners’ language aptitude has a major effect on their
success in learning an L2. As Gardner and MacIntyre (1992) comment:

Research makes it clear that in the long run language aptitude is probably
the single best predictor of achievement in a second language (1992: 215).

Horwitz’s and Skehan’s studies have helped to reopen the research agenda.
Skehan (1991) suggests three ways in which aptitude research might pro¬
gress. First, there is a need to revise the basic model to take account of current
theories of language and language learning. Second, the tests need to be
revised to more strongly reflect the kind of abilities involved in BICS. Third,
research needs to be conducted in a variety of learning contexts including
informal ones.
 Individual learner differences 499

In the light of Skehan’s conclusions, it is interesting to note that Carroll

himself is ‘somewhat skeptical about the possibilities for greatly improving
foreign language aptitude predictions beyond their present levels’ (1990: 27).
He specifically addresses the possibility of including tests of verbal fluency
factors—which might be considered of importance in communicative lan¬
guage teaching—but dismisses them on the grounds that these had been ex¬
plored in his early research and found not to make any significant
contribution to prediction. However, Carroll does offer a number of sugges¬
tions for improving aptitude measures. The existing tests might be ‘fine-
tuned’ (for example, by distinguishing auditory ability and phonemic coding
ability). Some new tests might be developed (for example, memory tests that
measure aspects of memory other than rote learning ability). He also suggests
that other potentially useful abilities for language learning might be identified
by examining the nature of the cognitive abilities needed to perform different
kinds of language learning tasks.
There have, in fact, been almost no attempts to develop alternative lan¬
guage aptitude tests. One exception is Parry and Child’s (1990) work on
VORD, an aptitude test based on an artificial language that was initially de¬
veloped in the early 1970s. The development consisted of incorporating a
component designed to test language aptitude in a contextual framework by
requiring subjects to apply general rules to particular segments of texts. How¬
ever, in a small-scale study involving 36 adult subjects, many of whom were
experienced language learners, Parry and Child found that MLAT was a bet¬
ter overall predictor of language-learning success than VORD. This study,
then, bears out Carroll’s skepticism about improving tests of language
aptitude.

Learning styles
The second general factor we will consider is learning style. The idea of learn¬
ing style comes from general psychology. It refers to the characteristic ways in
which individuals orientate to problem-solving. Keefe (1979) defines learning
style as:

... the characteristic cognitive, affective and physiological behaviours that

serve as relatively stable indicators of how learners perceive, interact with
and respond to the learning environment... Learning style is a consistent
way of functioning, that reflects underlying causes of behaviour.

Learning style, therefore, reflects ‘the totality of psychological functioning’

(Willing 1987). An individual’s learning style is viewed as relatively fixed and
not readily changed. However, Little and Singleton (1990) argue that it is
possible to help adult learners to explore their own preferences and to shape
their learning approach to suit the requirements of a particular learning task.
It is this belief that underlies the idea of ‘learner training’ (Holec 1987).
500 Explaining individual differences

A number of learning style distinctions have been made in cognitive psy¬

chology. Bruner, Goodnow, and Austin (1957) distinguished ‘focusers and
‘scanners’. The former tackle a problem by concentrating on one feature at a
time, in a step-by-step process, while the latter deal with several features at
the same time and allow their ideas to crystallize slowly. Pask and Scott
(1972) distinguished ‘serialists’ and ‘holists’, according to whether learners
operate with simple hypotheses (consisting of a single proposition) or com¬
plex hypotheses (involving multiple propositions). Other distinctions, which
seem to reflect personality as much as learning style differences, include ‘im¬
pulsive’ v. ‘reflective’ thinkers and ‘divergent’ v. ‘convergent’ thinkers. How¬
ever, only the distinction between field independence (FI) and field
dependence (FD) has attracted much attention in SLA research.

Field dependence/independence

This distinction is taken from the work of Witkin and associates. Witkin, Olt-
man, Raskin, and Karp (1971) provide the following description:

In a field-dependent mode of perceiving, perception is strongly dominated

by the overall organization of the surrounding field, and parts of the field
are experienced as ‘fused’. In a field-independent mode of perceiving, parts
of the field are experienced as discrete from organized ground ... ‘field de¬
pendent’ and ‘field independent’, like the designations ‘tall’ and ‘short’ are
relative (1971: 4).

To investigate subjects’ perception of a ‘field’ as ‘fused’ or as composed of

‘discrete parts’, a variety of tests of space-orientation (for example, the body-
adjustment test, which measured subjects’ perception of an upright position)
were developed. Later, Witkin and his associates developed various pencil-
and-paper tests that required subjects to locate a simple geometrical figure
within a more complex design. Witkin claimed that they measured the same
construct as the space orientation tests. The version most widely used in SLA
research is the Group Embedded Figures Test (GEFT).4 This provides a meas¬
ure of the extent to which individuals are field independent but, as McLaugh¬
lin (1985) has pointed out, there is no separate measure of field dependency.
As a result of research with the GEFT conducted by Witkin and his associ¬
ates, various claims regarding the relationship between FI/FD and other vari¬
ables have been advanced (see Table 11.4). Precisely what the test measures,
however, is a matter of some controversy. Griffiths and Sheen (1992) suggest
that many of Witkin’s claims are ‘expansive’. They argue that it is not possible
to extrapolate from visual-spatial abilities, which they claim the test meas¬
ures, to other cognitive abilities or to personality dimensions. They suggest
that the GEFT is a test of ‘ability’ rather than ‘style’, ‘specifically in the visuo-
spatial domain, but also related to general intelligence’ (1992: 141). This is a
view shared by Chapelle and Green (1992). They point to Witkin’s later work
 Individual learner differences 501

(for example, Witkin and Goodenough 1981), which defined FI/FD as involv¬
ing three major constructs: reliance on internal v. external referents, cognitive
restructuring skills, and interpersonal competencies. Flowever, they argue
that the GEFT only measures ‘cognitive restructuring ability’. They also sug¬
gest it constitutes a measure of one type of intelligence, ‘fluid ability’ (the abil¬
ity that is independent of any body of content knowledge and that is involved
in problem solving). The value of the L2 research based on the GEFT, then,
would seem to lie more in the light it throws on aptitude, as Chapelle and
Green’s (1992) subsequent discussion demonstrates, than in illuminating the
role of cognitive style. It is discussed in this section only because the bulk of
L2 studies have treated the GEFT as a measure of style.
Another problem with the GEFT is that it may be culturally biased,
favouring certain groups over others (see Willing 1987). Griffiths (1991b)
reports marked differences in the scores obtained by nationals of different
Asian countries (for example, 10 out of a maximum of 18 by Samoans as op¬
posed to 15 plus for Japanese). Certainly, the characterization of FI and FD
individuals, as shown in Table 11.4, is contentious.

Field independence Field dependence

adolescents/adults children
males females
object-oriented jobs people-oriented jobs
urban, technological societies rural, agrarian societies
free social structures rigid social structures
individualistic people group-centred people

Table 11 A: Variables associated with field in dependence and field dependence

There are now a considerable number of studies that have investigated the
relationship between FI/FD and L2 learning (see Table 11.5). One hypothesis
that has been investigated is that FI learners do better in formal language
learning, while FD learners do better in informal language learning. Flow¬
ever, with the exception of Abraham and Vann (1987), who studied only two
learners, this hypothesis has not received support. In general, FI learners do
better on measures of formal language learning (for example, discrete point
tests)—see, for example, studies by Seliger (1977), Stansfield and Hansen
(1983), Chapelle and Roberts (1986), and Carter (1988). But FI learners also
do better on integrative tests and tests of communicative competence, de¬
signed to favour FD learners—see studies by Hansen (1984), Chapelle and
Roberts (1986), and Carter (1988). Also, a number of studies (for example,
Bialystok and Frohlich (1978), Day (1984), and Ellis (1990b)) have failed to
find a significant relationship between GEFT scores and measures of learning.
Other studies comment on the weakness of the relationship. D’Anglejan and
502 Explaining individual differences

Renaud (1985), for instance, report that FI explained less than 1 per cent of
the variance in tests of all four language skills. Even some of the studies that
report a relationship between FI and L2 achievement also comment that it
loses significance once the effects of the learners’ general scholastic ability
have been statistically removed (for example, Hansen 1984). d’Anglejan and
Renaud (1985) found a considerable overlap between FI/FD, as measured by
the GEFT, and verbal intelligence, as measured by Raven’s Progressive Matri¬
ces, a result that led them to question Witkin and Berry’s (1975) insistence
that the two cognitive traits were distinct and that reinforces the views ex¬
pressed by Griffiths and Sheen (1992) and Chapelle and Green (1992) that
GEFT is really an aptitude test.
Another hypothesis which has not received convincing support from the re¬
search carried out to date is that FD learners will interact more and seek out
more contact with other users of the L2. Seliger (1977) found that FI learners
interacted more in the classroom. He argues that this was because they were
not reliant on the approval of others, and were therefore more prepared to
take risks, but his results could also be interpreted as contradicting the hypo¬
thesis that FD learners will interact more. Day (1984) found no relationship
whatsoever between FI/FD and participation. Carter (1988) reports that FI
learners were more concerned with meaning than FD learners, which also
works against the hypothesis.
Two studies have investigated learner-instruction matching. Abraham
(1985) reports that FI learners did better with a deductive method of instruc¬
tion, while FD learners benefited from being given examples. However, Car¬
ter (1988) found that FI learners did better than FD learners in both a formal
and a functional language course. Again, then, there is no clear support for
the hypothesis that FI learners gain from a form-focused, deductive approach
and FD learners from a meaning-focused, inductive approach.
A number of studies have also considered the FI/FD distinction in relation
to aptitude. Bialystok and Frohlich (1978) report a general relationship be¬
tween FI and aptitude (measured with MLAT). In Ellis 1990b I reported sig¬
nificant correlations of moderate strength (around .30) between aptitude
measures of grammaticality sensitivity, word memory and sound discrimina¬
tion, and FI. However, if it can be shown that the GEFT measures much the
same thing as aptitude tests, its value becomes questionable. As Skehan
(1991) points out, an advantage of aptitude tests is that they allow for learn¬
ers with both alternative orientations to be identified (for example, analytic v.
memory) and for learners with strong, combined orientations (for example,
analytic and memory). H. Brown (1987) has also suggested that some learn¬
ers may have ‘flexible’ cognitive styles, combining FI and FD modes of pro¬
cessing and adapting their approach to suit different learning tasks. However,
the GEFT is based on the assumption that the less one is FI, the more one is
FD, and so it cannot be used to investigate the presumed advantages of a flex¬
ible learning style.
 Individual learner differences 503

(icontinued overleaf)
^ c
>N 0 o 0 c ° c c
CO E 0 0
"O 0 0 n 0)
If 0) o
3 O
c
0
Q 3 0 0 E |
0 >
c

'? I 55 CO
CJ3 0
c \ 0
LL. C/5
0 g

0 0 o F © ' 0 ij8 c
^ a
^3 CO ^ E JO
E o 0 o
g CD o 1
0 P r-© JZ E
8« 75 o E o
0-° 0 0 IS 5 o
c
.c £
JZ u 0
o _ c- 0
0 b ll 0
NO 0 O
.§>8
0 x
I 0
z to
II 0 o
0s -H
jd m
0 o ^
<; o r
c
0
0 CD o ^ £ JZ . 0 0
c ® Q. 0
0 0 0 0
0 0 C -Q c -C o
J§ 0
O
TO E
~ o
£ c o
0
C
c
£ 3 ^ O £ 3 0 f|f
3 0
8 ®0
O &£ in -c
o cp
0
v_
1_
0
0
0—0
C LL 0
JZ
CD
#> s TO g ^ CD <<
2 E
0 u C OT 3
o £ 3 B O
O
■Q O)
0 g © c TO >
Soil,
2 o .9-8 5 0 0
JO CO TO .TO C 0 © 0 0
0 0 0
0 E 0 CD c g> ® f I £ N 0
3
^ 0
Q 0
b .E CD.P b £ 3 .P 0
— 0
0 3
© 0
©
o 5
(/> ° E O 1- 0 E ,-d c 0 0
0 0 O m CD 0 5 £ TO ■O
0
GC
JZ 0
° 0
LL 0
O 0
^ 0® 0 (0
“ E ©s E 0

CD C 0
C o 0
o JZ
0 — CO c
0 0 O J-T 0
‘3 U-> ®TO
TO
P k E 0 ^ O) e 0) 0)
0 ra © jf 3 ® c ■o I E 0 E
C C c c: 0 c c £
O O CO ® TO ~ o - o o CL to E 2 ® o £ E
0 0 .2
c c c
3
C CD *0 2 0 0 0 E ^ 0 *0 TO ra
±2^0
V) 0 C >,0 0)? o 55 o 8 8 C ® O c ^ 0
O 0 0O o £ c=
o)j= jz .y E IS k 0 0 -C 0 3 © TO > TO to
%m C 0 0 tt t >> U)
3 3 B
c > ®
TO O D 2 ® 55 © 0 0 3 co c 14- CD c0
if)
0
“ Q. Q. ^
0 0 0 2- to t> 9- ® 0 n CL ‘
| F CD TO C TO
0 0 ^
I,
N
0 0 3
o E E ■§ JZ c §f« 3 E N eI E E E E 0 0 0^)
O O E O 0 o -b c o o o 3 O 0 J- C
5 it C 0 o o ° E CD O O O .b 0 O
O O Q. 0 Q) -J O -b 3 0 ©

o 0
0 0
■*-> 0
0 0 0 0
0 0 CO
> 0 0 c 0
0) 0 0 6
^ 3 0
> E E 0) -o 0~
o 0 0 0 2 0 <r.)
c ■*-< Xm £
c C ‘go 0 0 0
0
v_ u. C\J c c
o 1° c c
*3 0 0 1 ?
CL 0 O 0 0 0 O) 'CD
O Q. 0)
i- o V- -O c 0 0
Q. 3 3 — 0 0 ^ 0 CD CD X

0 0 0 0 0 0 0
CD CD CD CD CD CD CD
0 0 0 0 0 0 0
3 3 3 3 3 3 3
CD CD CD CD CD CD CD
C C C C C C
0 JO JO JO JO JO
CD 0 0 0 0 c c 0
c C c c c g> CD C
o o o o 0 0 o
s o o o o i— o
0) 0 0 0 0 o o 0
C/) CO CO CO CO LL LL CO

3 0
0 i—
° 0 O CO 0 0
C\J u 0 3
0 0 CO —I 0 > >
0
'c o 'c
0 C £ 3 O (0
C 0 S § 3 0 3
0 C
0^ toO 0 o
0 0

o
0 C
ii>
0 0
0 c 0 C
Er
C
P CO
— v_ o 8 .© <
0 ° O -*-■ 0 — c CO 5-iSS
3 > o c 3 8 CO CO 3
!q “ 0 ° 5 3
3 o £ o £ in Q- in Q- r 0
5
< C
3 CO LL CO LL c\i CO .E csj CO .E <
CO C/) LL

■0 5 c
C 0 0
0 0
JZ C
00
C :0 0
0 O) I
0 CO
■Q CO CO
|fS C
CD
r^ C ■0 0
I? h- 0 c
G> 0 0
X TO
c
ZZ
o C
0 E
0
0 0 0 _ 0 To ^ JZ
>* (f) T-
■D CD E >% oo C co 0
3 S 00 0 CO
0 '0 0 CT) CD _Q
53 CO z 1« CO x- <
504 Explaining individual differences

0 (0 0
0 o > a

also related to greater variety of strategy use

N -Q o cd ,
_o ^ 0 ® H O ..u-
O 0 3 ® « CD
TJ T)
i-s

and to greater concern with correctness

0 O Id £ > 0
C CO CD CD . TJ w 2 2 15 "0
— o
LL O
II
CD _g ts
LU ®
TJ
0
>
0 £ E m .2 O
C CO o
a) cc
© £
CD
c ®
0 n
'0 .Q
O
0 ■g
life! TJ
0
2 i
D)-C 0 »_ E c H | 0
0 0 £ o tS0 0
-Q jz i 5 0 CO
co tc 0 CD 2 © O £ > O
co a> 3 n
.9-3 £ w 2 £
-C -Q ° <5 ° tj E 0
® P K 3
c 0 © 0 0
O) Q..® 3 0 V — ®||s§
O
£ § ~
TJ
g-ffl
W ■=
O
I £
0 Q
-Q LL £.2 § 2 LU
^ ,P ^ CD C 0 0 o
JS TJ TJ
0 <d
O = CD
O O Q. -C O
CO >- -
^ O » o
C -O
£ o c 3 tj 0
1 « co ~ -§"0
® d) c o co c E E p 0
§ | era® ■4= 0 0 o c 0
o ^ .2 ra E
—© 9> UJ
c 0 2 E C CL C C o 0
3 ~ o 0 t 0 0 0 £ e .S> c '-f-> k-
</> c~ t- ■gulf *“ TJ n © © © § 0 0 0 ^
a> .2> 0 (0 M- V O C u - _ CD x
®® 2
ec CO CL > O CD Z 0 0 o LL -E 0

Placement test; TOEFL;

o

test of spoken English;

(f) 0

different lang. learning

U)
c C 0 0 0
0 E o 0 0
‘E 0
i- E OJ’-P N
. - 0 0 0 0)0 o £ ®
0 o
i— • — O o o ~

performance on
0) >, 0
O 0 o
H— C 0 °-o) O)
o
0
0 £ „
E
ii
6
0
O
c
It
' © E H ©
c
LU
a> 1* (11 CL 2
c
k. O 0 ® CO o 0 o E 0
3 0)0 O)
0 0 E
Q-2 IIS

tasks
<0 N .£ o JZ
a) m ® o
O £ §■§ to E 0 0°
2 o O O) 0 & to o) 3 o

0
c
o
0 0
high and one low
0
> 0 0
0 TJ TJ
> 0 0 0 0
E
proficiency

O 0 0 E 0
c > k_ i-
0 0 tj 0 T 0
0 c
o C
■O E
0 -0
X 0 O)
c
o
± CD

0 0 0 0 0
CD O) CD O) CD
0 0 0 0 0
3 3 3 3
CD CD O) CD
c C c 0
_0 _0 _0 _0 _0
0) TJ TJ TJ TJ TJ
c C C 0 C C
o o o o o
5 o o o o g
0 0 0 0 0 0
(0 CO CO CO CO CO

l_- TJ 0
v-
o
C\| 0
s § o T o 0 0
d ^
o ±= E 0 .c 0 _ O
i
0 c 0
c
co tj ,2> 0 TJ © Q. i—
0
L2 English; LI

tj £ c e 0 E co 0
0 E c 0 3 8
E o _0
0 —
0
CD >>
i 0
TJ 0
_c
- '77'
.2
0

0 4= CO
0
—0 SZ
0 3
CD o E 0 2 C
0 0
— 0 0 TJ 3
0 O) o Q)
Spanish

Q 0 O 0 CL TJ
0 “ 0 ^
3 o) T 0 ■§ § o) E 3 3 CD C ^
C LO TJ £= 0 < 0
la 3 D) O
£ ■ £ "ra a>o 13 "S 0 LU CL -n o
3 ■g C 0 13 1- © C C 03 0 © t- CNJ © C
CO < LU CL O o
TJ .2 C£) _0 LU _0 _Q ^3 T,
CO J 3 0

T
0
.Q c
O 0
LO DC >
CO TJ TJ
CO CD c
CD 0
CO E _0
CD 0 0
>* J0
■D 0 Q- co
1987

3 > k_ © CO
0 0 J3 n
55 I a < 6? <
 Individual learner differences 505

0 4- .9-
0 CD
> 03
3
0 0 c
0 0
■O 8 o o _ E tc 0
O 0 3 O c
« CO
03 ~ » ® c O CL ~o o
C/5
0 fc
0 O E
r{.9 *+— c C
— 0
O ' I So « o O 0 00 "P
£
2? 3 o £ | in o J- 0
8 i 0 C O y-
E 03 _« a) "d a E
tf 03 05 3 » c .S' o) 8 $
0 C cs> => co j= .9
S* Q 1 £ Si o §
t
uj
*
c 2 0 0
-Q 0
o ®
s
$ °- ■O 0 E
3 |
0)0 .9 E
O
c >
0 C 0 03
0 2
■g n
£ §£ c0 0° 0. 0 .E
1 o
o 8 B E S
Q- 111 ■§ 8 c 0 0
_Q 0
•o .Q .£ -E o a> u
m § C ■c 0 m- -q
2 o c rr tj ^ C3
0 o
it Q. -n c a> ^ £ 0 CN
>% E
_0 ±L
0^P 0 0 Q 8
o 3 g
s § -J
3
0 -~ _0 O 0 S
<f.)
£ c 0 0
i_ i— “I2 ■E w
_0 0 C o
3 8 §
0 o 3 +* O 2^8 0 0 CD
cc 2 CT o E n: a s O O r
o oi cc E —
0
cu

o CD
o -o
c o ll g c
0
0
~o 3 0 LU
0
~o ^ .9 H- M. 0 N
O
E S
CO
3 O 0 ^o i_ &0
c s< 0 g' 0 M
£ co
CL 0 Set!' o
0 3 03 £ c 3 0 '+3 j: o c SZ to u
I S3 se
0
0
i_
3
0
£
®
|
®
3
CO
!|
jo g
3 £
> CO 05 —
§ ® £ c
~ 4- P 0
o
E
3 0
00 =;5-
8
4= 0 ^
C 3 p
0 .a g u
CO £ 3 0 0 0 E
0 5 <S I? 5>
® x
ra 0 O
O 03*= 2 o R8 ^
5 O E CC o uj t M £ =5 ^ 0 0 > CD
OJ
OJ
-c

.§)

V)
§2
> .S
0
c
0
C •S
c | E c
O)
0 EPis CD
0
m m .E CQ

.u
0 0 0
CD CD CD -5
0 0 0
3 3 5
05 CD
C C
_0 _0
O) C C C
c g> CD s

s '0 '0 0 </>

0 O
<0 o
^3
:; *ac
: 0 s
CL O *0
CN'“0 —
3
—1 ^ C 0
§■ °> c E
O CD
0 0 0 O
c C C
3 ° •
0 - o c 2 0 P I c 0 ..
c c T*H
o C I ^ ” f C -C 0 ~
J= .0 O CO o 0 0 T-H
0 0 c
!o 8 £ E 03 CL S'S E 0 _0 t ^33
3 5 §■ Q_ Q °
05
CO CO
-o
CO 05 CO .E 0 0 NCO o E (5
t2

r^ CO
CO co J2
05 05 o
y— CD
c i_ O)
■o o 0
tr
3 o
0 0
CO CQ o
506 Explaining individual differences

To conclude, the research into FI/FD has shed little light on the relationship
between cognitive style and L2 learning. Griffiths and Sheen (1992) are par¬
ticularly dismissive of it, arguing that the research is seriously flawed by a fail¬
ure to test hypotheses derivable from Witkin’s theory, a failure to recognize
that the GEFT measures ability rather than style, and a failure to acknow¬
ledge the rejection of Witkin’s theory in mainstream psychology. They con¬
clude that the FI/FD construct is ‘a wasteland, bereft of meaningful
hypotheses for T2 researchers’ (1992: 145). Quite apart from these theoret¬
ical objections, the research itself has proved inconclusive; most of the studies
have found either no relationship between FI and L2 achievement or only a
very weak one. If FI/FD and the GEFT have a future in SEA research, it is
probably in investigations of aptitude rather than of cognitive style. In this re¬
spect, as Chapelle (1992) points out in her response to Griffiths and Sheen,
the research may still have some value.

Other approaches to investigating learning styledn second

language learners
Some of the other approaches for investigating learning style are more prom¬
ising. A number of researchers have used survey techniques to collect data on
learners’ stated preferences. On the basis of such data, Reid (1987) distin¬
guished four perceptual learning modalities:

1 visual learning (for example, reading and studying charts)

2 auditory learning (for example, listening to lectures or to audio tapes)
3 kinaesthetic learning (involving physical responses)
4 tactile learning (hands-on learning, as in building models).

He then administered a questionnaire to 1,388 students of varying language

backgrounds to investigate their preferred modalities. This revealed that the
learners’ preferences often differed significantly from those of native speakers
of American English. They showed a general preference for kinaesthetic and
tactile learning styles (with the exception of the Japanese), and for individual
as opposed to group learning. With regard to the latter, Reid comments:
‘Every language background, including English, gave group work as a minor
or negative preference’. Proficiency level was not related to learning style
preference, but length of residence in the United States was—the longer the
period, the more an auditory style was preferred, reflecting perhaps an ad¬
aptation to the prevailing demands of the American educational system.
In another survey, Willing (1987) investigated the learning styles of 517 ad¬
ult ESL learners in Australia. Their responses to a 30-item questionnaire were
analysed by means of factor analysis (a statistical procedure designed to dis¬
cover if there were any combinations of items which afforded parallel re¬
sponses). Willing identified two major dimensions of learning style. One was
cognitive and corresponded closely to that of field independence/dependence.
 Individual learner differences 507

The other was more affective in nature; it concerned how active learners were
in the way they reported approaching L2 learning tasks. Skehan (1991) sug¬
gests that the second dimension reflects a personality as much as a learning
style factor. Based on these two dimensions, Willing describes four general
learning styles (summarized in Table 11.6). Willing’s study is interesting, but
it suffers from a number of methodological problems, which cast doubt on
the results obtained. Willing’s study did not include sufficient numbers of
subjects to permit a statistically valid analysis of many of the nationalities it
sampled. Also, the subjects were not equally distributed into the four learning
styles. Perhaps the most serious reservation concerns the validity and reliabil¬
ity of the questionnaire Willing used.

General learning style Main characteristics

1 Concrete learning style Direct means of processing information; people-

2 Analytical learning style
3 Communicative learning
style
4 Authority-orientated
learning style
orientated; spontaneous; imaginative; emotional; dislikes
routinized learning; prefers kinaesthetic modality.
Focuses on specific problems and proceeds by means of
hypothetical-deductive reasoning; object-orientated;
independent; dislikes failure; prefers logical, didactic
presentation. ""
Fairly independent; highly adaptable and flexible;
responsive to facts that do not fit; prefers social learning
and a communicative approach; enjoys taking decisions.
Reliant on other people; needs teacher’s directions and
explanations; likes a structured learning environment;
intolerant of facts that do not fit; prefers a sequential
progression; dislikes discovery learning.

Table 11.6: Four learning styles used by adult ESL learners (based on Willing
1987)

Willing’s questionnaire was also used (in a slightly adapted form) by Gieve
(1991) in a study of the learning styles of 156 first-year female students at a
Junior College in Japan. Gieve analysed the data using a variety of statistical
procedures, the most revealing of which was cluster analysis (a procedure
that groups people according to the similarity of their response profiles). Five
clusters emerged, which can be identified as follows: (1) learners with instru¬
mental motivation together with communicative orientation, (2) learners
with no motivation, (3) learners interested in general intellectual develop¬
ment, (4) learners with a strong motivation but with no clear aims, and (5)
learners with integrative motivation interested in living abroad. Most stu¬
dents fell into cluster (2). This analysis suggests that the strength and nature
of learners’ motivation works as a major dimension of learning style. Gieve
also makes the interesting point that the students’ responses reflected their
environment at least as much as innate qualities, as shown by the fact that
508 Explaining individual differences

significant differences were found in learners’ responses depending on

whether the questionnaire was administered by a Japanese or native-speaker
researcher! Again, then, doubts are raised as to whether the results obtained
by the questionnaire can be considered reliable.
Further evidence of differences in learning styles comes from product
analyses of learner language. Researching LI acquisition, Nelson (1973)
distinguishes ‘referential’ and ‘expressive’ learners. The former use language
to name things while the latter prefer to use it to indicate feelings, needs, and
social forms. Peters (1977) suggests that some learners are ‘analytic’ (i.e. are
word-learners and progress incrementally through a recognizable sequence
of stages of acquisition) and some are ‘gestalt’ (i.e. are sentence-learners who
begin with whole sentences which are used to perform functions that are
important to them). In SLA research, somewhat similar types of learners have
been identified. Hatch (1974), for example, talks of ‘rule-formers’, who pay
close attention to linguistic form, sort out the rules, and develop steadily, and
‘data-gatherers’, who show greater concern for communication and make
extensive use of formulaic chunks. Krashen (1978) distinguishes ‘monitor-
over-users’ and ‘monitor-under-users’, allowing also for ‘optimal monitor
users’. Dechert (1984a) compared the styles of two advanced learners in a
narrative reproduction task. One was ‘analytic’ (manifesting long pauses at
chunk boundaries, few corrections, and serial processing) while the other was
‘synthetic’ (manifesting shorter pauses throughout, more corrections, and
episodic processing). The terms used to characterize the differences in lan¬
guage produced by different learners proliferate, but it is tempting to identify
one general distinction that seems to underlie many of those mentioned
above—the experiential, communicatively-orientated learner as opposed to
the analytical, norm-orientated learner.
At the moment there are few general conclusions that can be drawn from
the research on learning style. Learners clearly differ enormously in their pre¬
ferred approach to L2 learning, but it is impossible to say which learning style
works best. Quite possibly it is learners who display flexibility who are most
successful, but there is no real evidence yet for such a conclusion. One of the
major problems is that the concept of ‘learning style’ is ill-defined, apparently
overlapping with other individual differences of both an affective and a cog¬
nitive nature. It is unlikely that much progress will be made until researchers
know what it is they want to measure.

Motivation

Language teachers readily acknowledge the importance of learners’ motiva¬

tion, not infrequently explaining their own sense of failure with reference to
their students’ lack of motivation. SLA research also views motivation as a
key factor in L2 learning. There have been differences, however, in the way in
 Individual learner differences 509

which teachers and researchers have typically conceptualized ‘motivation’

(see Crookes and Schmidt 1990).
In an attempt to characterize a non-theoretical view of motivation, Skehan
(1989) puts forward four hypotheses:

1 The Intrinsic Hypothesis: motivation derives from an inherent interest in

the learning tasks the learner is asked to perform.
2 The Resultative Hypothesis: learners who do well will persevere, those
who do not do well will be discouraged and try less hard.
3 The Internal Cause Hypothesis: the learner brings to the learning situation
a certain quantity of motivation as a given.
4 The Carrot and Stick Hypothesis: external influences and incentives will
affect the strength of the learner’s motivation.

These hypotheses have their correlates in the study of motivation in SLA re¬
search, but one of them, (3), has received the lion’s share of researchers’ atten¬
tion. We will begin, therefore, by examining the research which has addressed
this hypothesis.

Integrative motivation

According to Gardner’s socio-educational model (which we examined in

some detail in Chapter 6), an integrative orientation involves an interest in
learning an L2 because of ‘a sincere and personal interest in the people and
culture represented by the other language group’ (Lambert 1974: 98). It con¬
trasts with an instrumental orientation, which concerns ‘the practical value
and advantages of learning a new language’. ‘Orientation’, however, is not
the same as motivation, which is defined by Gardner as ‘the combination of
effort plus desire to achieve the goal of learning the language plus favourable
attitudes towards learning the language’ (1985: 10). Thus, whereas ‘orienta¬
tion’ refers to the underlying reasons for studying an L2, ‘motivation’ refers
to the directed effort individual learners make to learn the language. Over the
years, Gardner has become increasingly critical of research that focuses nar¬
rowly on the role of orientation in L2 learning, arguing that the effects of
learners’ orientations are mediated by their motivation—that is, whereas
orientation and L2 achievement are only indirectly related, motivation and
achievement are directly related.
What, then, is ‘integrative motivation’} Crookes and Schmidt (1989) point
out that the answer to this question is not entirely clear, as Gardner has pro¬
vided different operational definitions of this concept in different studies. As
we saw in Chapter 6, Gardner’s research rests on the use of self-report ques¬
tionnaires. Questions relating to a number of orientational and motivational
variables are included, and learners’ responses are factor-analysed in order to
identify general factors. One of the factors that has consistently emerged from
this procedure is ‘integrative motivation’. Table 11.7 shows the variables that
510 Explaining individual differences

Gardner and MacIntyre (1991) included in their composite measure of this

factor. It should be noted that the variable ‘integrative orientation’ is included
in this measure together with a number of other variables to do with the learn¬
ers’ attitudes, interest, and desire regarding the learning of French.

Variable Questionnaire items Example

Attitudes towards Five positively-worded and five ‘If Canada should lose the French
French Canadians negatively-worded items culture of Quebec, it would
indeed be a great loss’.
Interest in Foreign Five items expressing a positive ‘I enjoy meeting and listening to
Languages interest and five a relative people who speak other
disinterest. languages’.
Integrative Four items expressing the ‘Studying French can be
orientation importance of learning French for important because it allows
integrative reasons. people to participate more freely
in the activities of other cultural
groups’.
Attitudes towards Four items referring to French ‘French courses offer an excellent
the learning teachers in general and four opportunity for students to
situation referring to French courses--half broaden their cultural and
positive and half negative. linguistic horizons’.
Desire to learn Three positive and three negative ‘I wish I were fluent in French’.
French items.
Attitudes towards Three positive and three negative ‘I would really like to learn
learning French items. French’.

Table 11.7: An operational definition of ‘integrative motivation’ (based on

Gardner and MacIntyre 1991)

A consistent correlation between integrative motivation and L2 achieve¬

ment has been found in studies of anglophone Canadians learning French. In
the earlier research (for example, Gardner and Lambert 1972), integrative
orientation was seen as a more powerful predictor of achievement in formal
learning situations than instrumental orientation. In later research, Gardner
(1985) has continued to assert the importance of integrative motivation, al¬
though he now acknowledges that instrumental motivation can also lead to
successful learning (see below). However, Gardner argues that whereas in¬
strumental motivation emerges as a significant factor only in some studies,
integrative motivation has been found to be invariably related to L2
achievement.
Learners can, of course, have both integrative and instrumental motivation
(a possibility not clearly allowed for in the earlier work but acknowledged in
the later). Muchnick and Wolfe (1982), for instance, found that measures of
the integrative and instrumental motivation of 337 students of Spanish in
 Individual learner differences 511

high schools in the United States loaded on the same factor, suggesting that
for these learners, it was impossible to separate the two kinds of motivation.
Ely (1986b) investigated the types of motivation found in first-year university
students of Spanish in the United States. He found evidence of both strong
integrative and strong instrumental motivation. In this study, the two types
emerged as separate factors, but they were both present in the same students.
In order to demonstrate the overall effect of motivation on L2 achievement,
Gardner (1980; 1985) chooses to report the effects of a general measure of
motivation (based on the Attitude Motivation Index, which includes vari¬
ables relating to both integrative and instrumental motivation). A survey of
seven different geographical areas in Canada revealed a median correlation of
0.37 between the AMI scores and French grades. Thus, general motivation
(comprised primarily of measures of integrative motivation) accounts for ap¬
proximately 14 per cent of the variance in achievement scores. According to
Gardner, this constitutes a ‘remarkably strong’ relationship.
Some studies, however, have failed to find a positive relationship between
integrative motivation and L2 achievement. For example, Oiler, Baca, and
Vigil (1977) report that Mexican women in California who rated Anglo
people negatively were more successful in learning English than those who
rated them positively. Oiler and Perkins (1978) suggest that some learners
may be motivated to excel because of negative attitudes towards thextarget
language community. In this case negative feelings may lead to a desire to ma¬
nipulate and overcome the people of the target language—a phenomenon
which they refer to as Machiavellian Motivation. In other studies (for ex¬
ample, Chihara and Oiler 1978) the relationship between measures of integ¬
rative motivation and achievement has been weak and insignificant. Gardner
(1980) has defended the results of his own research vigorously, pointing out
the sheer number of studies which have reported a significant effect for integ¬
rative motivation and also attacking the design of the self-report question-
tpaires used in the studies by Oiler and associates.
There is something of a mismatch between the theoretical definition of mo¬
tivation which Gardner provides and the operational construct that he has in¬
vestigated. This is because'self-report questionnaires do not provide any
indication of the actual effort which learners put into their learning and
which Gardner sees as a key dimension of motivation. However, a number of
studies have tried to investigate ‘effort’ and a related concept, ‘persistence’,
seeking to show that measures of motivation (and, in particular, integrative
motivation) are related to the amount of effort and persistence that individual
learners display in classroom contexts. The assumption here is that ‘active’
learners will achieve more than ‘passive’ learners.
One possible definition of an active learner is one who participates fre¬
quently in classroom interaction. In a series of studies inspired by Gardner’s
socio-educational model, Gliksman (for example, in Gliksman 1976; Gliks-
man, Gardner, and Smythe 1982) has provided evidence to show that
512 Explaining individual differences

students studying French in Canadian high schools receive directed teacher

questions, volunteer answers, give correct answers, and receive positive re¬
inforcement according to the strength of their integrative motivation. The
higher their integrative motivation, the more these classroom behaviours are
evident. Naiman et al. (1978) found a similar relationship, although in this
case ‘effort’ on the part of the learners was also associated with instrumental
motivation. They were also able to show that a number of classroom behavi¬
ours associated with effort (such as hand-raising) were significantly related to
L2 achievement. However, the relationship between learner participation in
the classroom and L2 learning remains uncertain, as other studies have failed
to find one or have even come up with a negative one (see Chapter 14).
There have also been a number of studies that have investigated the rela¬
tionships between motivation, persistence, and achievement. Clement,
Smythe, and Gardner (1978) investigated what factors caused Canadian stu¬
dents in grades 9,10, and 11 to drop out of a French programme. They found
that motivation proved a more powerful predictor than language aptitude,
classroom anxiety, or even L2 achievement. Ramage (1990) investigated the
factors causing high school students of French and Spanish to drop out in the
United states. Her most significant finding was that students who chose to
continue their studies beyond the second year attached very low importance
to fulfilling curriculum requirements, and instead reported an interest in the
target-language cultures (which suggests an integrative motivation) and the
desire to attain proficiency in all language skills. There were other factors that
played a part, such as grades achieved in previous foreign language courses
and when the students started to study (the early starters were more likely to
continue). Ramage’s study suggests, not surprisingly, that, although integra¬
tive motivation contributes to persistence in foreign language study, it is not
the only factor.
Integrative motivation is not the only kind of internal motivation involved
in L2 learning, as Clement and his co-workers’ research demonstrates (see
Clement and Kruidenier 1983 and 1985; Clement 1986; Kruidenier and
Clement 1986). Clement (1986), for example, investigated 293 francophone
students at the University of Ottowa, dividing them into a majority and mi¬
nority group, depending on whether francophones were in the majority or
minority in their home areas. The subjects completed a questionnaire de¬
signed to produce measures of a variety of variables such as integrativeness,
fear of assimilation, motivation (defined in terms of ‘attitudes’ and ‘desire’),
self-confidence, frequency of contact, and acculturation. English language
ability was measured by means of an oral interview and a general proficiency
test. One of the main findings of this study was that integrativeness was not
related to language outcomes, nor was it influenced by the status of the learn¬
ers (i.e. whether they belonged to the majority or minority group). The best
predictor of language proficiency proved to be self-confidence, a finding that
held for both groups. Clement points out that many of Gardner’s earlier
 Individual learner differences 513

studies had investigated school students and that the rather different results
he obtained might reflect the greater maturity and autonomy of his
(Clement’s) subjects, who typically learned and used English on a day-to-day
basis, mainly outside the classroom. Clement suggests that for such learners
‘frequency of contact and the concomitant self-confidence might be more
important in determining second language proficiency than socio-contextual
or affective factors’ (1986: 287).
Kruidenier and Clement (1986) also failed to find any evidence in support
for Gardner’s integrative orientation in another study of language learners in
Quebec. Instead they found evidence of a number of different orientations
(friendship, travel, knowledge, and instrumental) with different groups of
learners revealing different dominant orientations, depending on their
learning situation. For example, learners of a minority language like Spanish
were more influenced by a travel orientation, whereas francophone learners
of English were more influenced by a friendship orientation. One possibility
raised by Dornyei (1990), however, is that these various orientations are all
part of a general integrative orientation, with different groups of learners
emphasizing different constituents of this.
To sum up, integrative motivation has been shown to be strongly related to
L2 achievement. It combines with instrumental motivation to serve as a
powerful predictor of success in formal contexts. Learners with integrative
motivation are more active in class and are less likely to drop out. However,
integrativeness is not always the main motivational factor in L2 learning;
some learners, such as those living in bilingual areas, may be more influenced
by other factors like self-confidence or friendship. There are also a number of
limitations to the research paradigm that has been used to study integrative
motivation (see Chapter 6). In particular, it takes no cognizance of the
potential effect that learning experiences can have on learners’ motivation, as
opposed to the effect that motivation has on language learning.

Instrumental motivation
The Carrot and Stick Hypothesis sees external incentives and influences as
determinants of learners’ motivational strength. It has been investigated in
SLA research through studies of instrumental motivation.
It is again useful to distinguish ‘orientation’ and ‘motivation’. Gardner and
MacIntyre (1991) measure the former by means of a self-report questionnaire
in which learners respond to statements such as ‘Studying French can be
important because it is useful for one’s career’. They equate ‘instrumental
motivation’ with giving students a financial reward for performing a task
successfully.
Much of the research has investigated the effects of an instrumental
orientation (as opposed to motivation) on learning. The results have been
variable, reflecting the situational/cultural context of learning. Thus, whereas
514 Explaining individual differences

instrumental motivation has been found to be only a weak predictor of for¬

eign language achievement in several Canadian studies (see Gardner and
Lambert 1972), it appears to be much more powerful in other contexts where
learners have little or no interest in the target-language culture and few or no
opportunities to interact with its members. For example, Gardner and Lam¬
bert (1972) found that a measure of instrumental orientation accounted for a
significant proportion of the variance in Tagalog learners of L2 English in the
Philippines. Similarly, Lukmani (1972) found that an instrumental orienta¬
tion was more important than an integrative orientation in non-westernized
female learners of L2 English in Bombay. The social situation helps to deter¬
mine both what kind of orientation learners have and what kind is most im¬
portant for language learning.
Surprisingly, there have been few studies which have investigated the direct
effect of an instrumental motivation (i.e. the provision of some kind of incent¬
ive to learn). In a very early study, Dunkel (1948) (cited in Gardner and
MacIntyre 1992) offered financial rewards to students learning Farsi and
found that although this did not result in a significantly better performance
on a grammar test, there was a tendency in this direction. Gardner and
MacIntyre (1991) report a study in which 46 university psychology students
were rewarded with $10 if they succeeded in a paired-associate (English-
French) vocabulary task, while the same number were just told to do their
best. The students offered the reward did significantly better. They also spent
more time viewing the pairs of words, except on the sixth and last trial in the
task, when the possibility of a reward no longer existed. This led Gardner and
MacIntyre to claim that once any chance for receiving a reward is eliminated,
learners may cease applying extra effort. They see this as a major disadvant¬
age of instrumental motivation.
To sum up, learners with an instrumental reason for learning an L2 can be
successful. In some ‘second’ as opposed to ‘foreign’ settings an instrumental
orientation may be the most important one. Providing learners with incent¬
ives (such as money) may also aid learning by increasing the time learners
spend studying, but the effects may cease as soon as the reward stops.

Resultative motivation

In the studies considered above, integrative and instrumental motivation

were seen as causes of L2 achievement. However, the majority of studies were
correlational in design and, as has often been pointed it, it is not possible to at¬
tribute direction in a correlational relationship. We do not know, therefore,
whether motivation is to be seen as the cause or the result of success in L2
learning or both.
Gardner (1985) claims that motivation constitutes a causative variable, al¬
though he is also prepared to accept that some modification of learners’ at¬
titudes can arise as a result of positive learning experiences, particularly in
 Individual learner differences 515

courses of a short duration. Spolsky (1989) reviews a number studies (for ex¬
ample, Gardner, Smythe, and Brunet 1977; Gardner, Smythe, and Clement
1979) which suggest that ‘while greater motivation and attitudes lead to bet¬
ter learning, the converse is not true’ (1989: 153).
Other studies, however, suggest that learners’ motivation is strongly affec¬
ted by their achievement. Strong (1983; 1984) investigated motivation and
English language attainment in Spanish-speaking kindergarten children and
found that fluency in English preceded an inclination to associate with target-
language groups. Savignon (1972) reported that students’ desire to learn
French increased with gains in French proficiency. Finally, a study by Her-
mann (1980) also suggested that it is success that contributes to motivation
rather than vice-versa. Hermann advanced the ‘Resultative Hypothesis’,
which claims that learners who do well are more likely to develop motiva¬
tional intensity and to be active in the classroom.
The Resultative Hypothesis may be particularly applicable in contexts
where learners have very low initial motivation. Berwick and Ross (1989) in¬
vestigated 90 first-year Japanese university students majoring in international
commerce and taking obligatory English classes. These students had a strong
instrumental motivation to learn the English they needed to pass the univer¬
sity entrance examinations, but typically became demotivated once they were
at university. Berwick and Ross found little evidence of any motivation on a
pre-test administered at the beginning of the English course, but much more
on a post-test given at the end. The students’ motivation appeared to broaden
as a result of the course with two new motivational factors, labelled ‘support’
and ‘interest’, emerging. This study provides clear evidence of ‘an experiential
dimension to learner’s motivation’.
It is likely that the relationship between motivation and achievement is an
interactive one. A high level of motivation does stimulate learning, but per¬
ceived success in achieving L2 goals can help to maintain existing motivation
and even create new types. Conversely, a vicious circle of low motivation =
low achievement = lower motivation can develop.

Motivation as intrinsic interest

The notion of intrinsic motivation is an old one in psychology. It was de¬
veloped as an alternative to goal-directed theories of motivation that emphas¬
ize the role of extrinsic rewards and punishments. Keller (1984) (cited in
Crookes and Schmidt 1989) identifies ‘interest’ as one of the main elements of
motivation, defining it as a positive response to stimuli based on existing cog¬
nitive structures in such a way that learners’ curiosity is aroused and sus¬
tained. It is this view that underlies discussions of motivation in language
pedagogy. Crookes and Schmidt (1989: 16) observe that ‘it is probably fair to
say that teachers would describe a student as motivated if s/he becomes pro¬
ductively engaged on learning tasks, and sustains that engagement, without
516 Explaining individual differences

the need for continual encouragement or direction’. Teachers see it as their

job to motivate students by engaging their interest in classroom activities.
One way in which intrinsic interest in L2 learning might be achieved is by
providing opportunities for communication. McNamara (1973) has argued
that ‘the really important part of motivation lies in the act of communica¬
tion’. Rossier (1975) also emphasizes the importance of a desire to commun¬
icate, arguing that without this an integrative motivation may not be
effective. It is the need to get meanings across and the pleasure experienced
when this is achieved that provides the motivation to learn an L2.
One possibility, supported by a strong pedagogic literature (see Holec
1980 and 1987; Dickinson 1987), is that interest is engendered if learners be¬
come self-directed (i.e. are able to determine their own learning objectives,
choose their own ways of achieving these, and evaluate their own progress).
Dickinson refers to a study by Bachman (1964) which indicated that involv¬
ing learners in decision-making tended to lead to increased motivation and,
thereby, to increased productivity. Gardner, Ginsberg, and Smythe (1976)
compared the effects of two sorts of instructional programmes on 25 learners
of French at Dalhousie University in Canada. One was very tradi¬
tional—lockstep teaching with a heavy focus on grammatical accur¬
acy—while the other was innovative, with individualized instruction and
opportunity for free communication. The students who experienced the tra¬
ditional programme indicated that they were more likely to withdraw and
had a more negative view of their French teacher. Those who experienced the
innovatory programme reported a greater desire to excel and a more positive
attitude to learning French. Diary studies also provide evidence to suggest
that self-direction is important to learners. For example, J. Schumann (Schu¬
mann and Schumann 1977) reports frustration with the instructional pro¬
gramme he was following because of his wish to maintain a personal
language learning agenda. It is possible, however, that not all learners will
benefit from opportunities for self-direction (see the earlier section on learn¬
ing style).
Crookes and Schmidt (1989) suggest a number of other ways in which
teachers seek to foster intrinsic motivation. They try to make sure that the
learning tasks pose a reasonable challenge to the students—neither too diffi¬
cult nor too easy. They provide opportunities for group work. They base
tasks on their perceptions of learners’ needs and wants and they try to provide
for plenty of variety in classroom activities. Above all, perhaps, they try to en¬
sure that motivation is engendered as a result of a good rapport with the
learners. As Finocchiaro (1981) puts it:

Motivation is the feeling nurtured primarily by the classroom teacher in the

learning situation. The moment of truth—the enhancement of motiva¬
tion—occurs when the teacher closes the classroom door, greets his stu¬
dents with a warm, welcoming smile, and proceeds to interact with various
 Individual learner differences 517

individuals by making comments or asking questions which indicate per¬

sonal concern.

However, there has been very little systematic research of the effects these
various pedagogic procedures have on motivation.

The effect of motivation

Motivation in L2 learning constitutes one of the most fully researched areas

of individual differences. The bulk of the research, however, has focused
rather narrowly on integrative and instrumental motivation, relying almost
exclusively on self-report questionnaires and correlational designs. Little
work on motivation as intrinsic interest has taken place. Also, little attention
has been paid to the effect of motivation on the process of learning (as op¬
posed to the product). Crookes and Schmidt (1989) argue that research that
links teachers’ and learners’ actions to persistence and effort in language
learning would have a more ‘real-world’ impact. Skehan (1989; 1991) argues
for more research on motivation in naturalistic as opposed to classroom
settings.

Personality

In the eyes of many language teachers, the personality of their students consti¬
tutes a major factor contributing to success or failure in language learning.
Griffiths (1991b), for example, conducted a survey of 98 teachers of ESL/EFL
in England, Japan, and Oman in order to determine how important they rated
personality and two other IDs. He reports a mean rating of 4 on a five-point
scale—slightly higher than the rating for intelligence and just below that for
memory. Learners also consider personality factors to be important. Of the
‘good language learners’ investigated by Naiman et al. (1978) 31 per cent be¬
lieved that extroversion was helpful in acquiring oral skills. It is somewhat
surprising, therefore, to find that the research that has investigated personal¬
ity variables and L2 learning is so scanty and, in many ways, so un¬
satisfactory.
This research is broadly summarized in Table 11.8. A number of general
observations arise from this summary. First, the personality variables consti¬
tute a very mixed bag. Some relate to well-established theories of personality
(such as extroversion/introversion) but have been investigated without refer¬
ence to the theory from which they have been drawn. Others are based only
very loosely on constructs in general psychology (for example, risk-taking).
Others entail an extension of a psychological construct to make it applicable
to L2 learning (for example, Guiora’s notion of a ‘language ego’).
A second observation, evident from the descriptions of the different dimen¬
sions of personality in Table 11.8, is that the variables are sometimes vague
and overlap in ill-defined ways. The concepts of extroversion, empathy, and
518 Explaining individual differences

TD
0
0 >% O
CO 0 ■O c ■O >* I s E
£ to 0 c o >N-Q 0 O O 3
3 3 c .-t^ O 0 JC
to
11 -Q 0 0 C ■£ 0 ^ ~o 0 0
0 _0 ... o O 0 CO
i— 0 o 'O 0- 0 .9
5 i E
> O C 0 0 0 I ? § “ d O >
-p o
XL o) o O 0 -C
i— +-' 0 ^ CL 0 >3 © C 0

0 —1 c c
0
> c *® CD O o _q .t: 3
3 0 C © 4/5 o £ 0
nsz
tSs
O D-
0 0 C ^ C
55-2 ■OOO
II
0 0
C -C O 0
O 5 C c to tr - 0
— Q_ 0
3
o CLjz
0 ‘ '
^ ^ '0 f?T3 § C O
o — 0 o _ > •- 0
0 0 0 c
0.0 d
5|«|
05 C o
0 0 £ 0 -o Z
f- c 0
E
^ 0 Q. ^ © 0
c o © c §■ C 0 ^
> 3 O
05 :C ©
3 -O o **- c
3 § !c o £ 0 3 s5
cn o 0 c t
.g co -p
c .2>
I SI 8
|l ill
O
H—
0
4-< t 0
O 0 O
c/> 2%
©
i- C © © 0 § .g 0 £ ■O 2 iS 0 0 ~ 0 0 i5 b 2 o

c 5o •.2 o § 0 Q.
O si 8 "O
>>
© -O ~
b T3 c c
.© 0 C
-9 Y -O
0 11 ra E I&8| •i fc m ® © o 2 0 .g> 3 0 0
© 3
2 CQ
0 cr z o £ 8 SOc a -5 p 8 8 © 0 I o Q. —I 0 (ODD

0
CO O
CO
05 '3 ■o
0 O 0 c
O) C 0
c CO ■D CM
h- i_
o s| 0 r-
v_ 05 05
? (T & CM
CO
co
0 0'
uc 0
CM
b CO 0 N-
o 0 0 © CD 0
3 CD 0 “ t
+* C 05
(/) CO 0 T~ ® p 0
a> 0 ©
C o >. E •|S
0 •i ©S | 55 p „
‘to 3
2 CQ zo? Z 0

0 (-
0 .E
05
o 3 ©
0 c
c CD _0 0 ©
0 0 CD C
0 c 0 0
E O © g
_ 0 - o o
3 CO 3 T3. . =5
i- 0 0 >. 0 to
c 0 C
0 •- 0 0 55 .£ p
(/) o ^ p p £ c 3
C 3 0 0
0 Q_ cr « E cr
0 LU SUs
D) Y c c E Y
C 8 t E oO O Q O
o Ills
'C
3
£ ^
a o Q.
0
S2 R
0 0
LU ^0-P
0 CL
X
0
CL
0
0 0 O 0 0 —
0 ® 1 Q. p -3 0 E 2 2
0
0 0 9 S-a X 0 0 -2 I
"O 3 7=
> £ 0 ,5 © 0
III c CQ 0 X ^ LU CL CO < 0 cr.E
2 CO

0
0 c
O 0
© fit' ~o
035 R-© "0
C m!
00^
■K 0>Y
0 © © C
> • O O © 05 Q. 3 0
-43 0 ■*-* 1 4- * ■*-• f“ R 0
0 CO
o 0 3 ;
0 S « ° £ SS? 0 © 00
3 £ ® 03 0 0 v
:§ ® 3 « ® =E « 0 o’ n o
«„ t> £ ^ C 0 N 0 0 o c 0
0 02 E ® 3 ~
©- rr
O C §2o ■g 3
2? t5 ©
■° > 8
0 > c 0
II §-§ © 0
— o i ® c CD 0 to ® c R Yj o 0
- 0 5 CL O c ^ > P Q.'O
•0 o 5 p | o 0
c 0 o JZ ’c 3 tO O
E S-’g j3 T3
>; £ | 0^0) 2 © o . 05 c
II '0 0 0 sp
0 0 c0 «
R -C 3 3 -a 0 p
c © 5
Y 0 aJ c P O C 3 ® p 3 O
^ -C 0 e- i
0 © "O jz 0 E
S £ O 2 o 5 Y C £ 0
*- Q. P p 0 ‘1 ^ >3 Xl 0
3 0 0 0 -2
_0 "O ~ 0 w © _
0 E m 05 > TJ
0 0 irt 5- E
« ™ © - S2 o 0 to 0 o © 0 © > 0 —1
O 0 “ © 0 "O 0 0 © ^ -
0
O ?o 5s>®
-c
.■? o © — c
c 0
© E £ Q. O TJ
5 ® c O -C 0
3 © '© | 00© O ^ CO © • >
0 CD '- c 0 0
0^0
sZ 0
© “ >»
110
^
0 0 0 0
0 _
0 E
0 — -X
0-2

o £•£
O

8 a
0 © ^ t; 0 o <D £ *0 0 E «Q
or « ® »- X5 0 0 r.2
O Y Q- := ^ O?
0 II -* E © c d: •- »o2S .0 "O .0 TD '0 o c
0 0 o c/5 © •- k r: c E c JZ C 0 C R 05 o
D CC 8 - ^ £ I- 03 lu o H © 0 0 Y © o

c 05 0 E
0 C 0 >> 0
> 0 c
C/5 It: 0 ±: 0
L C 3 -C
0 0 to
o 0 05 to 0 '•*->
0 > 1 n
O 0 !q a
a if) j=
0 b E E *0
C be h- 0 LU CO —
<
 Individual learner differences 519

risk-taking would appear to be related, as would those of self-esteem and in¬

hibition, but in what ways is not clear.
Third, the instruments which have been used to measure (or control) the
personality variables are varied and, in some cases, of doubtful validity and
reliability. Most of them consist of self-report questionnaires of different
kinds (for example, the Eysenck Personality Inventory and the Hogan Em¬
pathy Scale). Some of them (such as the ones just mentioned) have been de¬
veloped by psychologists and have been subjected to extensive testing.
Others, however, have been devised specifically for use with L2 learners (such
as Ely’s and Gardner and Lambert’s questionnaires), and it is not always clear
that these instruments measure what they purport to measure. This is perhaps
most true in the series of studies conducted by Guiora and associates.
In one set of studies (Guiora, Lane, and Bosworth 1967; Taylor, Guiora,
Catford, and Lane 1969; Guiora et al. 1972) an instrument known as the
Micro-Momentary Expression test was used to measure learners’ empathy by
having them identify when a woman in a film changed her facial expression.
In the first of the studies a positive correlation between empathy (so meas¬
ured) and accuracy of pronunciation by fourteen French teachers was re¬
ported. A subsequent study, however, found no such relationship, while a
third study found a positive relationship for some languages and a negative
one for others. In another set of studies (Guiora et al. 1972; Guiora, Acton,
Erard, and Strickland 1980) learners were administered quantities of alcohol
and valium in order to reduce their level of inhibition. Alcohol resulted in bet¬
ter pronunciation but valium did not—a result that has interesting possibil¬
ities for application in language teaching! However, as H. Brown (1987) has
pointed out, even the positive results obtained for alcohol cannot be inter¬
preted as showing that reduced inhibition facilitates L2 pronunciation, as it is
possible that the effects of the alcohol were due to reduced muscular tension
(a purely physical phenomenon). While not all the studies mentioned in Table
11.8 are of such doubtful validity as Guiora’s, instrumentation is a general
problem.
A fourth observation based on Table 11.8 is that the results obtained by
different studies are inconsistent. Thus, whereas Naiman et al. (1978) found a
positive relationship between tolerance of ambiguity and scores on a listening
comprehension test, they failed to find any relationship with scores on an im¬
itation test. Chapelle and Roberts (1986) also report low correlations be¬
tween this personality factor and criterion measures of L2 proficiency.
Similar inconsistencies can be observed in the main results for extroversion,
empathy, and self-esteem. One explanation is that different instruments were
used to measure the same personality variable, but this is not always the case
(as can be seen in Naiman et al.’s results). The main problem is that there is
often no theoretical basis for predicting which personality variables will be
positively or negatively related to which aspects of L2 proficiency.
520 Explaining individual differences

It is easy to take a rather bleak view of personality studies in SLA research.

There follows a detailed discussion of the research that has examined one as¬
pect of personality—extroversion/introversion—which shows that the study
of personality does hold considerable promise.

Extroversion/introversion
The extroversion/introversion distinction refers to one of several dimensions
or traits which together constitute an individual’s personality (see Eysenck
1970; Cattell 1956). In Eysenck’s theory, for example, there are three major
dimensions—extroversion/introversion, neurotic/stable and psychotic/nor¬
mal. In choosing to investigate the extroversion/introversion distinction,
therefore, researchers are investigating only one aspect of learners’
personality.5
Extroversion/introversion represents a continuum (i.e. individuals can be
more or less extroverted), but it is possible to identify idealized types:

Extraverts are sociable, like parties, have many friends and need excite¬
ment; they are sensation-seekers and risk-takers, like practical jokes and
are lively and active. Conversely introverts are quiet, prefer reading to
meeting people, have few but close friends and usually avoid excitement.
(Eysenck and Chan 1982: 154)

The extent to which individuals verge towards one of these types is usually
measured by analysing responses to self-report questions such as those in the
Eysenck Personality Questionnaire.
There are two major hypotheses regarding the relationship between extro¬
version/introversion and L2 learning. The first—which has been the most
widely researched—is that extroverted learners will do better in acquiring ba¬
sic interpersonal communication skills (BICS). The rationale for this hypo¬
thesis is that sociability (an essential feature of extroversion) will result in
more opportunities to practise, more input, and more success in communicat¬
ing in the L2. The second hypothesis is that introverted learners will do better
at developing cognitive academic language ability (CALP). The rationale for
this hypothesis comes from studies which show that introverted learners typ¬
ically enjoy more academic success, perhaps because they spend more time
reading and writing (see Griffiths 1991b).
There is some support for the first hypothesis. Strong (1983) reviewed the
results of 12 studies which had investigated extroversion or similar traits
(sociability, empathy, outgoingness, and popularity). He shows that in the 8
studies where the criterion measure was ‘natural communicative language’
and which, therefore, provided an indication of BICS, 6 of them showed that
extroversion was an advantage. Strong then reports his own study of 13
Spanish-speaking kindergarten children in which various dimensions of per¬
sonality were investigated using both classroom observation and Coan and
 Individual learner differences 521

Cattell’s (1966) Early School Personality Questionnaire. Interestingly, he

found that those measures derived from observation accounted for nearly all
the statistically significant correlations with language measures based on the
children’s natural communicative language. The important variables were
talkativeness, responsiveness, and gregariousness, all of which are behavi¬
ours associated with extroversion.
The second hypothesis has received less support. Strong’s survey of studies
that have investigated the effects of introversion on ‘linguistic task language’
reveals that less than half report a significant relationship. Busch (1982), in a
study not included in Strong’s survey, also failed to find a relationship. She
used the Eysenck Personality Inventory to obtain measures of extroversion/
introversion. Her subjects were adolescent and adult Japanese learners of
English in Japan. The correlations between extroversion and scores on a four-
part written proficiency test were non-significant (with one exception) and
generally negative in the case of the adolescents, and non-significant but pos¬
itive in the case of the adults. This study, then, also fails to lend much support
to the hypothesis that introversion aids the development of academic lan¬
guage skills. Interestingly, Busch also failed to find a clear relationship be¬
tween extroversion and scores on an oral interview test (presumably more
representative of BICS). It is possible that extroversion/introversion has little
to do with L2 learning in Japan because the classroom situation requires all
learners to exhibit introverted-type behaviour out of respect for the teacher.
One profitable line of enquiry might be to investigate the relations between
personality variables and the different kinds of behaviours that learners en¬
gage in. Ehrman (1990) examined the relationship between personality and
choice of learning strategies in a study we will consider in the next chapter.
Robson (1992) undertook a small-scale study of Japanese college students,
measuring personality by means of the Yatabe/Guilford Personality Invent¬
ory, and obtaining measures of voluntary participation from oral English
classes. His general finding was that learners who are extrovert and emotion¬
ally stable were more likely to engage in oral participation than introverts and
neurotics. It does not follow, of course, that such learners will be more suc¬
cessful, as high levels of classroom participation may not enhance L2
learning.
To sum up, the evidence linking extroversion to the acquisition of BICS is
fairly substantial, but there is no clear support for the claim that introversion
benefits the acquisition of CALP. Furthermore, the effects of extroversion/
introversion may be situation-dependent, evident in some learning contexts
but not in others. Also, there is a suggestion (in Strong’s research) that better
results may be obtained when the measures of extroversion/introversion are
based on observation than when they are based on self-report questionnaires.
Finally, it should be noted that better results may be obtained if researchers
explore specific interactions between personality measures such as
522 Explaining individual differences

extroversion/introversion and learners’ performance under different instruc¬

tional conditions, as suggested by Griffiths (1991b).

Summary
The following statements constitute a general summary of the research re¬
viewed in the previous sections. Readers are also referred to the summary
statements at the end of each section in this chapter.

1 Beliefs

Learners have been found to hold different beliefs about how an L2 is best
learnt. These differences reflect their past learning experiences and general
factors such as learning style and personality.

2 Affective state

Learners’ affective states vary dynamically and have a significant impact on

their ability to learn. Anxiety arising out of poor performance, communica¬
tion apprehension, tests, and fear of negative evaluation is likely to have a
debilitating effect on L2 learning, but it can also have a facilitative effect.
How anxiety affects learning will depend on its strength and the situational
context.

3 Age

Children generally enjoy an advantage over adults in L2 learning because of

their age, particularly in pronunciation. However, this will only become evid¬
ent after substantial exposure to the L2. In the short term, adults may learn
faster. The evidence relating to the existence of a critical period for L2 ac¬
quisition, after which full competence is not possible, is mixed, with no defin¬
ite conclusion possible. Children and adults manifest similar processes of
learning.

4 Aptitude

Both quantitative and qualitative differences in language aptitude have been

found. These relate to the development of both linguistic and communicative
L2 abilities. Language aptitude involves both an underlying language learn¬
ing capacity and a capacity to handle decontextualized language. Language
aptitude has been found to be one of the best predictors of L2 learning.

5 Learning style

Learners manifest different learning styles (for example, analytical v.

experiential) but it is not yet clear whether some styles result in faster and
more learning than others. The distinction between field dependent and field
 Individual learner differences 523

independent, which has attracted considerable interest in SLA research, is

best treated as a distinction of ability rather than style. In many studies, it
does not emerge as a major factor in L2 learning.

6 Motivation

Strength of motivation serves as a powerful predictor of L2 achievement, but

may itself by the result of previous learning experiences. Learners with either
integrative or instrumental motivation, or a mixture of both, will manifest
greater effort and perseverance in learning. Other internal sources of motiva¬
tion, such as self-confidence, may be more important than either type of mo¬
tivation in some contexts. Motivation can also take the form of intrinsic
interest in specific learning activities and, as such, may be more easily influ¬
enced by teachers than goal-directed motivation.

7 Personality

The relationship between personality variables and L2 learning is not yet

clear. There is some evidence to show that extroverted learners are advant¬
aged in the development of the kind of language associated with basic inter¬
personal communication skills. Extroverted learners may also be more likely
to participate actively in oral communication.
These individual differences produce variation in the rate of learning and
the ultimate level of L2 attainment. Overall, there is little evidence to suggest
that they have a marked effect on the internal processes that account for in¬
terlanguage development.

Conclusion
As Skehan (1991) has pointed out there is still no comprehensive theory of
IDs in SLA research. A full theory will need to identify those IDs that are im¬
portant for successful learning, indicate the relative contribution of particular
IDs to learning, specify how IDs interrelate, account for their influence on the
learner’s choice of specific strategies, and account for the effect that learning
outcomes can have on IDs. It will also have to make clear what effect (if any)
IDs have on the process of L2 acquisition.
One way of viewing IDs is as conditions of learning (Spolsky 1989). In gen¬
eral, IDs constitute ‘typical’ and ‘graded’ rather than ‘necessary’ conditions.
For example, learners do not have to have a high aptitude or a particular
learning style or one type of personality in order to learn an L2. Rather, some
learners do better than others because they satisfy the conditions for learning
relating to these IDs to a greater extent. Some of the factors, however, might
be viewed as necessary conditions. For example, it may be necessary for
learners to start learning an F2 before puberty if they are to acquire a native
524 Explaining individual differences

pronunciation. Similarly, learners probably need at least a modicum of

motivation to learn anything.
Research into IDs to date has told us little about the relative strength of dif¬
ferent learner factors or how they interrelate. Gardner’s research has shown
that aptitude and motivation constitute separate factors, both accounting for
a substantial amount of the variance in learners’ L2 proficiency. These factors
have explanatory power. However, the picture is less certain where many of
the other factors (in particular learning style and personality) are concerned.
This is largely because the various constructs that have been investigated have
not been defined with sufficient rigour and have been poorly operationalized.
These factors do not have much explanatory power at the moment.
Two views regarding the nature of the relationship between IDs and L2
learning are possible. According to the aggregate view, success is the result of
the accumulative effect of facilitative IDS. For example, a child with low anxi¬
ety, high overall language aptitude, an inclination to be analytic, strong integ¬
rative motivation, and an outgoing personality could be considered likely to
succeed. According to the alternative view, there are many ways to achieve
success and it is not possible to draw up a single profile of the successful learn¬
er. This view of IDs has important implications for language instruction be¬
cause it acknowledges the need to take account of learner-instruction
interactions (i.e. to recognize that different learners can achieve the same level
of success if the instruction matches their own preferred approach to learn¬
ing). Learner-treatment interactions will be examined in Chapter 14.
As Skehan (1991) notes, ID research has been primarily concatenative in
approach. It has been mainly concerned with scale and test construction and
has employed statistical techniques to identify the interrelationships among
the multiple influences on learning and their relationship to L2 achievement.
This approach has a number of problems, not least the question of how best
to measure ID and criterion variables. Although Skehan considers the ‘fertil¬
ity’ of the concatenative approach ‘far from exhausted’ (1991: 292), he also
argues for more studies that employ a naturalistic methodology. Such studies
can shed light on the individuality of single learners and can also show the
dynamic nature of the interaction between the more malleable aspects of indi¬
vidual difference (for example, anxiety and motivation) and learners’ learn¬
ing experiences. A further advantage of naturalistic studies is that they can
provide measures of learning based on natural language use rather than on
tests or ratings, a major need in ID research.

Notes

1 A study by Castagnaro (1992) has attempted to use physiological measures

of learners’ classroom anxiety and, interestingly, found sizeable and signi¬
ficant positive correlations between these measures and measures of anxi¬
ety obtained from anxiety questionnaires.
 Individual learner differences 525

2 One of the interesting results obtained by Coppieters was that the differ¬
ences between the native-speaker and non-native-speaker judgements were
not uniform. Coppieters reports that divergence was less marked in con¬
structions ‘normally covered by the term UG’ and more marked in ‘func¬
tional’ distinctions (for example, passe compose v. imparfait). This result,
then, provides support for the claim that adult L2 learners have continued
access to UG (see Chapter 10). It should be noted, however, that Birdsong’s
(1992) replication of Coppieter’s study failed to find that native-speaker/
non-native-speaker divergences were explicable in terms of UG and
non-UG constructions.
3 Ioup (1989) also investigated an L2 learner who began L2 learning as a
child but still failed to acquire native-speaker ability despite very favour¬
able learning conditions. However, in this case the learner did acquire a
native accent, failing only to acquire certain grammatical and semantic dis¬
tinctions. Ioup finds this surprising, commenting that ‘one would assume
that the ability to acquire the phonology would imply the ability to acquire
syntax and semantics’ (1989: 171). It is possible, however, that her sub¬
ject’s difficulty with grammar and semantics reflects the general difficulty
she experienced where cognitive academic language ability (CALP) is con¬
cerned—a difficulty that many native speakers may also experience.
4 There is also the Embedded Figures Test (EFT) and the Child Embedded
Figures Test (CEFT). Different versions of these tests also exist. One of the
problems claimed by Griffiths and Sheen (1991) is the lack of equivalence
of these different versions.
5 There are, in fact, several approaches to describing and investigating per¬
sonality in the psychological literature. Whereas Eysenck’s trait theory has
been influential in Britain, factor analytic models (for example, Cattell
1956) have been popular in the United States. Robson (1993), in a review
of these (and other) theories, claims that the traits of neuroticism and ex¬
troversion are the ‘most unquestionably basic’ in that they figure in all
approaches.

Further reading

Surveys of individual differences in second language learners

Skehan’s surveys of the research into individual differences are comprehens¬
ive, objective, and very readable:
P. Skehan, Individual Differences in Second-Language Learning (Edward
Arnold, 1989).
P. Skehan, ‘Individual differences in second language learning’. Studies in Se¬
cond Language Acquisition (1991) 13: 275-98.
526 Explaining individual differences

Another very readable but more subjective account of many of the particular
individual differences discussed in this chapter can be found in:
H. D. Brown, Principles of Language Learning and Teaching (2nd Edition)
(Prentice Hall, 1987).
A useful overview of quantitative research is:
R. Gardner, ‘Second Language Learning in Adults: Correlates of Proficiency’.
Applied Language Learning (1991) 2: 1-29.

Suggested readings for particular IDs

Beliefs
E. Horwitz., ‘Surveying student beliefs about language learning’ in A. Wenden
and J. Rubin (eds.): Learning Strategies in Language Learning (Prentice Hall,
1987).

Affective factors (anxiety)

K. Bailey, ‘Competitiveness and anxiety in adult second language learning:
looking at and through the diary studies’ in H. Seliger and M. Long (eds.):
Classroom Oriented Research in Second Language Acquisition (Newbury
House, 1983).
E. Horwitz and D. Young, Language Learning Anxiety: Prom Theory and
Research to Classroom Implications (Prentice Hall, 1991).

Age
There are a number of excellent books dealing with the age issue:
D. Singleton, Language Acquisition: The Age Lactor (Multilingual Matters,
1989).
T. Scovel, A Time to Speak: a Psycholinguistic Inquiry into the Critical Period
for Human Speech (Newbury House, 1988).
B. Harley, Age in Second Language Acquisition (Multilingual Matters,
1986).
In addition, Long provides a thoughtful and provocative discussion of the
Critical Period Hypothesis:
M. Long, ‘Maturational constraints on language development’. Studies in
Second Language Acquisition (1990) 12: 251-85.
There are many individual studies of the age issue. A recent very interesting
one is
D. Birdsong, ‘Ultimate attainment in second language acquisition’. Language
(1992): 68: 706-55.
 Individual learner differences 527

Aptitude

A number of important articles on language aptitude can be found in:

K. Diller (ed.), Individual Differences and Universals in Language Aptitude,
(see in particular the articles by Carroll and Wesche) (Newbury House
1981).
Also recommended is:
J. Carroll, ‘Cognitive abilities in foreign language aptitude: then and now’ in
T. Parry and C. Stansfield (eds.): Language Aptitude Reconsidered (Prentice
Hall Regents, 1990).

Learning style

A useful general account of learning styles, together with an account of one

study, can be found in:
K. Willing, Learning Styles in Adult Migrant Education (Adult Migrant Edu¬
cation Resource Centre: National Curriculum Resource Centre, 1988).
Perhaps a good way into the wealth of literature on FI/FD in SFA research is
Griffiths and Sheen’s polemical critique:
R. Griffiths and R. Sheen, ‘Disembedded figures in the landscape: a re¬
appraisal of F2 research on field dependence-independence’. Applied Lin¬
guistics {1992) 13: 133-48.
C. Chapelle’s response to it is in Applied Linguistics (1992) 13: 375-84.

Motivation

A good summary of Gardner’s work and views on motivation can be found

in:
R. Gardner, Social Psychology and Second Language Learning: The Role of
Attitudes and Motivation (Edward Arnold 1985).
Numerous articles on motivation can be found in Language Learning, includ¬
ing the important:
G. Crookes and R. Schmidt, ‘Motivation: reopening the research agenda’.
University of Hawaii Working Papers in ESL (1989) 8: 217-56.

Personality
The literature is less impressive on personality, but Strong provides a useful
survey of the early research together with an account of his own study:
M. Strong, ‘Social styles and the second language acquisition of Spanish¬
speaking kindergartners’. TESOE Quarterly (1983) 17: 241-58.
'
12 Learning strategies

Introduction

In the last chapter we investigated individual differences by looking at the re¬

lationship between various learner factors and learning outcomes. However,
little was said about the mechanisms that establish these relationships. In this
chapter we will consider the mediating role of learning strategies. According
to the model of L2 acquisition shown in Figure 12.1, individual learner differ¬
ences (beliefs, affective states, general factors, and previous learning experi¬
ences) together with various situational factors (the target language being
studied, whether the setting is formal or informal, the nature of the instruc¬
tion, and the specific tasks learners are asked to perform) determine the learn¬
ers’ choice of learning strategies. These then influence two aspects of learning:
the rate of acquisition and the ultimate level of achievement. The success that
learners experience and their level of L2 proficiency can also affect their
choice of strategies.
The study of learning strategies has seen an ‘explosion of activity’ in recent
years (Skehan 1991: 285). In this chapter we will examine the main findings
of this research. We will begin by considering a number of definitions of
‘learning strategy’ and then consider the main methods that have been used to
investigate them. This will be followed by a discussion of various frameworks
for classifying learning strategies. The next section will deal with the factors
that influence learners’ choice of strategies. The relationship between learn¬
ing strategies and L2 learning is considered with reference to studies of the
‘good language learner’, cross-sectional studies that try to identify correla¬
tions between strategy use and L2 proficiency, and longitudinal studies of
groups of learners. Finally, attempts to train learners to employ particular
strategies will be considered.

Defining ‘learning strategies’

The concept of ‘strategy’ is a somewhat fuzzy one, and, as we noted in the in¬
troduction to Part Four, not easy to tie down. There we settled for a general
definition; a strategy consisted of mental or behavioural activity related to
some specific stage in the overall process of language acquisition or language
use. We will now examine a little more fully what a ‘strategy’ consists of in an
attempt to arrive at a definition of ‘learning strategy’.
530 Explaining individual differences

Figure 12.1: The relationship between individual learner differences,

situational factors, learning strategies, and learning outcomes

A distinction is often made between three types of strategies: production,

communication, and learning. Tarone (1980b: 419) considers the first two of
these under the heading ‘Strategies of language use’. A production strategy
consists of ‘an attempt to use one’s linguistic system efficiently and clearly,
with a minimum of effort’. Examples are simplification, rehearsal, and dis¬
course planning. Communication strategies, which we examined in some de¬
tail in Chapter 9, consist of attempts to deal with problems of communication
that have arisen in interaction. A language learning strategy is ‘an attempt to
develop linguistic and sociolinguistic competence in the target language’. Ex¬
amples given by Tarone include memorization, initiation of conversation
with native speakers, and inferencing. These distinctions are important, but
as Tarone observes, they are not easily applied as they rest on learners’ inten¬
tions which are often not clear or easy to establish. There is, for example, no
easy way of telling whether a strategy is motivated by a desire to learn or a de¬
sire to communicate.
It is also useful to distinguish two types of learning strategy: language
learning strategies and skill learning strategies. The former, as defined by
Tarone, are concerned with the learners’ attempts to master new linguistic
and sociolinguistic information about the target language. The latter are con¬
cerned with the learners’ attempts to become skilled listeners, speakers, read¬
ers, or writers. Writers, for instance, employ a variety of strategies for
exploiting teachers’ feedback on their written compositions (see Cohen 1987;
1991). Again, however, the distinction is not an easy one and the literature on
learning strategies does not always distinguish clearly between these two
 Learning strategies 531

Source Definition

Stern 1983 ‘In our view strategy is best reserved for general tendencies or overall
characteristics of the approach employed by the language learner, leaving
techniques as the term to refer to particular forms of observable learning
behaviour.’
Weinstein and ‘Learning strategies are the behaviours and thoughts that a learner
Mayer 1986 engages in during learning that are intended to influence the learner’s
encoding process.’
Chamot 1987 ‘Learning strategies are techniques, approaches or deliberate actions that
students take in order to facilitate the learning, recall of both linguistic and
content area information.’
Rubin 1987 ‘Learning strategies are strategies which contribute to the development of
the language system which the learner constructs and affect learning
directly.’
Oxford 1989 ‘Language learning strategies are behaviours or actions which learners use
to make language learning more successful, self-directed and enjoyable.’

Table 12.1: Definitions of learning strategies

types. Cohen (1990: 15), for instance, refers to learning strategies directed at
the ‘language skill’ of ‘vocabulary learning’, although this is clearly an aspect
of linguistic knowledge. The focus of this chapter will be on language learning
strategies.
A sample of definitions of language learning strategies taken from the re¬
cent literature (see Table 12.1) reveals a number of problems. It is not clear
whether they are to be perceived of as behavioural (and, therefore, observ¬
able) or as mental, or as both. Oxford (1989) appears to see them as essen¬
tially behavioural, whereas Weinstein and Mayer (1986) see them as both
behavioural and mental.
A second problem concerns the precise nature of the behaviours that are to
count as learning strategies. Here there is considerable uncertainty. Stern
(1983) distinguishes ‘strategies’ and ‘techniques’. The former are defined as
general and more or less deliberate ‘approaches’ to learning (for example, ‘an
active task approach’), whereas the latter constitute observable forms of lan¬
guage learning behaviour evident in particular areas of language learning
such as grammar (for example, ‘inferring grammar rules from texts’) and vo¬
cabulary (for example, ‘using a dictionary when necessary’). Other re¬
searchers, however, have used the term ‘strategy’ to refer to the kind of
behaviours Stern calls techniques.
A third problem is whether learning strategies are to be seen as conscious
and intentional or as subconscious. Many of the definitions in Table 12.1
avoid addressing this issue, but Chamot (1987) refers to them as ‘deliberate
actions’. Seliger (1984) distinguishes ‘strategies’ and ‘tactics’. He defines the
former as ‘basic abstract categories of processing by which information
532 Explaining individual differences

perceived in the outside world is organized and categorized into cognitive

structures as part of a conceptual network’ (1984: 4). In contrast, ‘tactics’ are
variable and idiosyncratic learning activities, which learners use to organize a
learning situation, respond to the learning environment, or cope with input
and output demands. This distinction is helpful. It is clear that what Seliger
refers to as ‘strategies’ was what we were focusing on in Chapter 9 when we
examined various cognitive theories of L2 learning. What he calls ‘tactics’ is
the focus of the present chapter. However, as we will shortly see, not all re¬
searchers make such a clear distinction between ‘strategies’ and ‘tactics’ on
the basis of consciousness. Some consider that what starts out as a conscious
‘tactic’ may evolve into a subconscious ‘strategy’. Also, useful as it might be to
make a terminological distinction along the lines proposed by Seliger, L2
acquisition researchers have not done so, using the term ‘strategy’ to refer to
both conscious and subconscious activities. We will follow suit in this chap¬
ter, although our focus will be on learning strategies as conscious or at least
potentially conscious actions which learners employ intentionally (i.e.
Seliger’s ‘tactics’).
A fourth problem concerns whether learning strategies are seen as having a
direct or an indirect effect on interlanguage development. Rubin (1987),
somewhat controversially, asserts that the effect is a direct one. But other re¬
searchers, such as Seliger, consider it to be more indirect—strategy use pro¬
vides learners with data, upon which the ‘deep’ subconscious processes can
work.
Finally, there are differences in opinions about what motivates the use of
learning strategies. All the definitions in Table 12.1 recognize that they are
used in an effort to learn something about the L2, but Oxford (1989) also sug¬
gests that their use can have an affective purpose (i.e. to increase enjoyment).
Perhaps one of the best approaches to defining learning strategies is to try
to list their main characteristics. The following list characterizes how the term
‘strategies’ has been used in the studies to be considered in this chapter:

1 Strategies refer to both general approaches and specific actions or tech¬

niques used to learn an L2.
2 Strategies are problem-orientated—the learner deploys a strategy to over¬
come some particular learning problem.
3 Learners are generally aware of the strategies they use and can identify
what they consist of if they are asked to pay attention to what they are
doing/thinking.
4 Strategies involve linguistic behaviour (such as requesting the name of an
object) and non-linguistic (such as pointing at an object so as to be told its
name).
5 Linguistic strategies can be performed in the LI and in the L2.
6 Some strategies are behavioural while others are mental. Thus some strat¬
egies are directly observable, while others are not.
 Learning strategies 533

7 In the main, strategies contribute indirectly to learning by providing

learners with data about the L2 which they can then process. However,
some strategies may also contribute directly (for example, memorization
strategies directed at specific lexical items or grammatical rules).
8 Strategy use varies considerably as a result of both the kind of task the
learner is engaged in and individual learner preferences.

Definitions of learning strategies have tended to be ad hoc and atheoretical.

However, O’Malley, Chamot, and Walker (1987) and O’Malley and Chamot
(1990) have attempted to ground the study of learning strategies within the
information-processing model of learning developed by Anderson (1980;
1983), which we considered briefly in Chapter 9. We noted that Anderson
distinguishes three stages of skill-learning: (1) the cognitive stage, where the
learner is involved in conscious activity resulting in declarative knowledge,
(2) the associative stage, where the learner strengthens the connections
among the various elements or components of the skill and constructs more
efficient production sets, and (3) the automatic stage, where execution be¬
comes more or less autonomous and subconscious. Anderson’s theory pro¬
vides for two interpretations of the term ‘strategy’. One, favoured by
Rabinowitz and Chi (1987, cited in O’Malley and Chamot 1990) is that strat¬
egies only occur in the early cognitive stage when they are conscious; they
cease to be ‘strategic’ when they are performed automatically. The other view
is that strategies occur in all three stages of development. They take the form
of production sets (i.e. ‘if ... then’ statements). For example, the strategy of
inferencing has this form:

If the goal is to comprehend an oral or written text, and I am unable to

identify a word’s meaning, then I will try to infer the meaning from context.

Initially, such sets exist only in declarative form; they are conscious and can
only be accessed through controlled processing. Gradually, they are proced-
uralized, until a point is reached where the learner is no longer conscious of
employing them. This is the view that Chamot and O’Malley seem to hold.
However, this difference in view may not be of much significance as strategies
can only be effectively studied in the declarative stage of learning, when learn¬
ers are able to verbalize them. For research purposes, therefore, strategies can
be defined as production sets that exist as declarative knowledge and are used
to solve some learning problem.

Methods used to investigate learning strategies

Attempts have been made to identify different learning strategies by observ¬
ing learners performing a variety of tasks, usually in classroom settings. Some
researchers have found this approach to be ‘not very productive’ (Rubin
1981), since it reveals nothing about the mental operations learners use and,
frequently, classrooms afford little opportunity for learners to exercise
534 Explaining individual differences

behavioural strategies. Naiman et al. (1978) and Cohen and Aphek (1981)
also comment on the failure of classroom observation to provide much in¬
formation about learners’ strategies. Chesterfield and Chesterfield (1985),
however, were able to report on a number of strategies that young children
used in a bilingual classroom. It may be, therefore, that observation works
with young children, whose behaviour may serve as a good indicator of their
mental activity, but not with adult learners, who often engage in internal pro¬
cessing not linked to actual behaviour.
A method that has been found to be more successful involves the use of
structured interviews and questionnaires, both of which call for retrospective
accounts of the strategies learners employ. A large number of studies have
used these methods (for example, Naiman et al. 1978; Rubin 1981; Politzer
and McGroarty 1985; Oxford 1985; Wenden 1986a; Chamot 1987). Inter¬
views and questionnaires can require learners to report on the learning strat¬
egies they use in general or in relation to a specific activity. For example,
Naiman et al. (1978:108) asked fairly general questions about the techniques
learners used for studying the sound system, grammar, vocabulary, and the
four skills:

Would you say that you have developed any language habits (gimmicks,
tricks, ways and techniques) that you would find useful in learning a new
language?
1) in learning the sound system
e.g. reading aloud to yourself (in front of a mirror), repeating words silently
to yourself after the teacher etc.

Wenden (1987a) conducted interviews in which she asked learners to com¬

ment on specific learning activities. She had them first complete a ‘Grid of
daily activities’ and then asked them to ‘recreate’ each activity by describing
it. Finally, she asked more specific questions relating to the strategies they
used to express themselves, to understand what was said to them, and to
think in the L2. Methods such as these have provided the most detailed
information about learning strategies, although Rubin (1981) notes that
le'arners vary greatly in their ability to describe their strategies and that
most of the students she investigated needed to be tutored in self-reporting.
Diary studies serve as another way of collecting information on learning
strategies by means of retrospective reports. As we saw in Chapter 11, the di¬
ary studies have been primarily used to shed light on the learners’ affective
states and how these influence learning. Some researchers, however, have
used them to explore learning strategies (for example, Brown 1985; Parkin¬
son and Howell-Richardson 1990).
Think-aloud tasks that require concurrent reporting have also been used
profitably (for example, by Abraham and Vann 1987; Mangubhai 1991).
These require learners to introspect on the strategies they employ while per¬
forming a particular task. In general, think-aloud tasks may tell us more
 Learning strategies 535

about skill learning (such as translation) than language learning strategies

(see articles in Fasrch and Kasper 1987), although Mangubhai’s study of the
reception strategies used by students when they responded to oral commands
shows that they can also shed light on the use of learning strategies.
It is not always practical to use think-aloud tasks. Beginner learners will
only be able describe the strategies they use in their LI, and this may interfere
with the way in which they perform the learning task. Also, learners may vary
in their ability to think aloud as they perform the task, with the result that the
strategies uncovered by such research may reflect the learning activities of a
rather special group of learners. There are, however, other ways in which
learners can be induced to verbalize their strategies.
One way is to ask learners to work in pairs to discuss their solutions to par¬
ticular tasks. Haastrup (1987) has used this technique to good effect in a
study of lexical inferencing strategies used by Danish learners of English. She
found that ‘pair thinking aloud’ provided ‘rich and informative data’ (1987:
202), although she points to the problem of ‘enormous variability between
protocols’, suggesting that the learners’ ability to verbalize their strategies
was still an issue. Morren (1993) also used pair thinking aloud, but allowed
his Japanese subjects to discuss in their LI, a procedure that worked well with
these learners, who possessed little oral ability in English.
Many of the most successful studies have employed multiple data collec¬
tion procedures (for example, retrospective accounts obtained through inter¬
views together with introspective accounts of performance on specific tasks).
As O’Malley and Chamot (1990:95) point out, a disconcerting feature of this
approach is that the strategies identified by each procedure often vary
considerably.
Many of the earlier studies of learner strategies (for example, Naiman et al.
1978) were entirely exploratory, aiming at identifying and describing a broad
range of strategies. More recent studies (for example, O’Malley 1987; Mang-
hubai 1991) have tended to focus on the strategies used to learn while per¬
forming individual tasks.

Classifying learning strategies

Much of the earlier research (for example, Rubin 1975 and 1981; Wong-
Fillmore 1976 and 1979; Stern 1975; Naiman et al. 1978) concentrated on
compiling inventories of the learning strategies that learners were observed to
use or reported using. Little attempt was made to classify the strategies into
general categories. The strategies identified tended to reflect the type of learn¬
ers under study, the setting, and the particular interests of the researchers.
Summarizing this early work, Skehan (1989) identifies three areas common
to the different taxonomies. One is ‘the learner’s capacity to impose himself
on the learning situation’. Naiman et al. refer to ‘an active task approach’,
which is evident in such behaviours as seeking out and responding positively
536 Explaining individual differences

to learning opportunities and engaging in practice activities. ‘Clarification/

verification’, which Rubin (1981) puts at the top of her list of strategies, also
belongs to this area. This strategy includes asking for examples of how to use
a word or expression, putting words in a sentence to check understanding,
looking up words in a dictionary, and paraphrasing a sentence to check
understanding. Wong-Fillmore lists two strategies that fall into this area:
‘get some expressions’ and ‘make the most of what you have got’.
The second general area concerns the learner’s ‘technical predispositions’.
Examples are Naiman et al.’s ‘realization of language as a system’ (which in¬
volves making judicious use of crosslingual comparisons, analysing the target
language, and making inferences about it), Rubin’s ‘guessing/inductive in-
ferencing’, and Wong-Fillmore’s ‘look for recurring parts in formulae’.
The third general area involves the learner’s capacity to evaluate. For ex¬
ample, Naiman et al. and Rubin both refer to the importance of monitoring.
This involves testing out guesses, correcting errors, and noting the sources of
errors.
Flowever, the early taxonomies differ in a number of ways, reflecting the
particular subjects that the researchers worked with. Thus Rubin and Nai¬
man et al., who elicited information from adults, emphasize the importance
of learners reflecting on their own learning and of conscious analysis, while
Wong-Fillmore, who studied 5-7-year-old Spanish-speaking children in play
situations, emphasizes the social aspects of learning. For example, Rubin lists
‘memorization’ and ‘deductive reasoning’ among her strategies, while
Wong-Fillmore identifies ‘join a group’ and ‘count on your friends’ as
important. These differences also seem to reflect whether the setting is a
formal or informal one, raising the possibility that the learning strategies
involved in classroom and naturalistic acquisition may not be the same.
Subsequent descriptive studies have endeavoured to identify broad classes
of learning strategies, under which large numbers of more specific strategies
can be grouped. The work of O’Malley and Chamot (for example, O’Malley
et al. 1985a and 1985b; O’Malley and Chamot 1990), Wenden (1991), and
Oxford (1990) has made an important contribution to our knowledge of
learning strategies.
In O’Malley and Chamot’s framework (see Table 12.2, pages 537-8), three
major types of strategy are distinguished, in accordance with the informa¬
tion-processing model, on which their research is based. Cognitive strategies
refer to ‘the steps or operations used in problem-solving that require direct
analysis, transformation or synthesis of learning materials’ (Rubin 1987).
They have an operative or cognitive-processing function. Among the cognit¬
ive strategies listed by Chamot (1987) are ‘repetition’ (imitating a language
model, either covertly or overtly), ‘note-taking’ (writing down information
presented orally), and ‘elaboration’ (relating new concepts to other informa¬
tion in memory). Cognitive strategies such as these appear to be directly
linked to the performance of particular learning tasks. Metacognitive
 Learning strategies 537

Learning Strategy Description

Metacognitive
Advance organizers Making a general but comprehensive preview of the concept or
principle in an anticipated learning activity.
Directed attention Deciding in advance to attend in general to a learning task and to
ignore irrelevant distractors.
Selective attention Deciding in advance to attend to specific aspects of language input
or situational details that will cue the retention of language input.
Self-management Understanding the conditions that help one learn and arranging for
the presence of those conditions.
Advance Planning for and rehearsing linguistic components necessary to carry
preparation out an upcoming language task.
Self-monitoring Correcting one’s speech for accuracy in pronunciation, grammar,
vocabulary, or for appropriateness related to the setting or to the
people who are present.
Delayed production Consciously deciding to postpone speaking to learn initially through
listening comprehension.
Self-evaluation Checking the outcomes of one’s own language learning against an
internal measure of completeness and accuracy.
Cognitive
Repetition Imitating a language model, including overt practice and silent
rehearsal.
Resourcing Defining or expanding a definition of a word or concept through use
of target language reference materials.
Directed physical Relating new information to physical actions, as with directives.
response
Translation Using the first language as a base for understanding and/or
producing the second language.
Grouping Reordering or reclassifying and perhaps labelling the material to be
learned based on common attributes.
Note-taking Writing down the main idea, important points, outline, or summary of
information presented orally or in writing.
Deduction Consciously applying rules to produce or understand the second
language.
Recombination Constructing a meaningful sentence or larger language sequence by
combining known elements in a new way.
Imagery Relating new information to visual concepts in memory via familiar
easily retrievable visualizations, phrases, or locations.
Auditory Retention of the sound or similar sound for a word, phrase, or longer
representation language sequence.
538 Explaining individual differences

Learning Strategy Description

Key word Remembering a new word in the second language by (1) identifying a
familiar word in the first language that sounds like or otherwise
resembles the new word, and (2) generating easily recalled images of
some relationship with the new word.
Contextualization Placing a word or phrase in a meaningful language sequence.
Elaboration Relating new information to other concepts in memory.
Transfer Using previously acquired linguistic and/or conceptual knowledge to
facilitate a new language learning task.
Inferencing Using available information to guess meanings of new items, predict
outcomes, or fill in missing information.
Social/affective
Cooperation Working with one or more peers to obtain feedback, pool
information, or model a language activity.
Question for Asking a teacher or other native speaker for repetition, paraphrasing,
clarification explanation and/or examples.

Table 12.2: O’Malley and Chamot’s typology of learning strategies (Chamot

1987)

strategies make use of knowledge about cognitive processes and constitute an

attempt to regulate language learning by means of planning, monitoring, and
evaluating. They have an executive function. Examples given by Chamot are
‘directed attention’ (deciding in advance to pay attention to specific aspects of
language input) and ‘self-management’ (displaying understanding of the con¬
ditions which help learning and trying to bring these about). Chamot gives
examples from interviews with ESL learners of the kinds of behaviours that
are representative of these strategies. Self-management, for instance, is evid¬
ent in the following comments from learners:

I sit in the front of the class so I can see the teacher’s face clearly.

It’s a good idea to mix with non-Hispanics because you’re forced to prac¬
tise your English. If you talk with a Chinese who is also studying English
you have to practise the language because it is the only way to
communicate.

Social/affective strategies concern the ways in which learners elect to interact

with other learners and native speakers. Chamot gives as examples ‘co¬
operation’ (‘working with one or more peers to obtain feedback, pool infor¬
mation or model a language activity’) and ‘question for clarification’ (‘asking
a teacher or other native speaker for repetition, paraphrase, explanation
and/or examples’).
Wenden’s (1983) research examined the strategies that adult foreign lan¬
guage learners use in order to direct their own learning. Eler focus, therefore,
 Learning strategies 539

is on what O’Malley and Chamot call metacognitive strategies. She identifies

three general categories of self-directing strategies: (1) knowing about lan¬
guage (relating to what language and language learning involves), (2) plan¬
ning (relating to the what and how of language learning), and (3)
self-evaluation (relating to progress in learning and the learner’s response to
the learning experience). Wenden found that adult learners pose questions
relating to each category and then take decisions depending on the kind of an¬
swers they come up with. For example, in the planning category, learners
might ask ‘What should I learn and how?’ and then go on to decide on lin¬
guistic objectives, resources, and use of resources. Wenden’s framework was
devised as a basis for learner training.
Perhaps the most comprehensive classification of learning strategies to date
is that provided by Oxford. Oxford built on the earlier classifications with the
aim of subsuming within her taxonomy virtually every strategy previously
mentioned in the literature. The classification scheme she first came up with
(Oxford 1985) was used as a basis for constructing a questionnaire on learn¬
ing strategies. The Strategy Inventory for Language Learning (SILL) (Oxford
1986) contained items tapping sixty-four individual strategies divided into
two main groups, primary strategies and support strategies (which corres¬
pond closely to O’Malley and Chamot’s cognitive and metacognitive types).
The inventory has undergone considerable revision since, as has the tax¬
onomy of strategies.
In Oxford (1990) a new taxonomy is presented, the general framework of
which is shown in Figure 12.2. A general distinction is drawn between direct
and indirect strategies. The former consist of ‘strategies that directly involve
the target language’ in the sense that they ‘require mental processing of the
language’ (1990: 37), while the latter ‘provide indirect support for language
learning through focusing, planning, evaluating, seeking opportunities, con¬
trolling anxiety, increasing cooperation and empathy and other means’
(1990: 151). The subcategories of direct and indirect strategies shown in Fig¬
ure 12.2 have familiar labels. Each subcategory is broken down into two fur¬
ther levels. For example, one type of cognitive strategy is ‘practising’, which
consists of five different kinds of behaviour (repeating, formally practising,
recognizing and using formulas, recombining, and practising natural-
istically). The scheme is marred by a failure to make a clear distinction be¬
tween strategies directed at learning the L2 and those directed at using it.
Thus, somewhat confusingly, ‘compensation strategies’ are classified as a dir¬
ect type of ‘learning strategy’. In this Oxford departs from other researchers,
who treat compensation strategies as distinct from learning strategies (for ex¬
ample, Rubin 1987). However, the organization of specific strategies into a
hierarchy of levels and the breadth of the taxonomy is impressive.
Considerable progress has been made in classifying learning strategies.
From the early beginnings, when researchers did little more than list
strategies, comprehensive, multi-levelled, and theoretically-motivated
540 Explaining individual differences

I. Memory strategies

Direct
II. Cognitive strategies
strategies

III. Compensation strategies

Learning
strategies

I. Metacognitive strategies

Indirect
II. Affective strategies
strategies

III. Social strategies

Figure 12.2: Diagram of a strategy system: Overview (from Oxford 1990:16)
taxonomies have been developed. The frameworks developed by O’Malley
and Chamot, Wenden, and (in particular) Oxford, provide a basis for study¬
ing which strategies or combinations of strategies are effective in promoting
learning (see the later section on ‘Learning strategies and language learning’,
page 545). Problems still remain, however. The categories that have been es¬
tablished are ‘high-inference’ in nature, their identification often requiring
considerable interpretation on the part of the researcher. The strategies listed
as belonging to a single type frequently vary on a number of dimensions such
as specificity (for example, ‘repetition’ is much more specific than ‘self¬
management’) and the extent to which they are observable (for example,
‘question for clarification’ constitutes an overt behaviour, while ‘elaboration’
does not). It is not yet clear whether the range of strategies available to the
learner is finite or infinite in number. These problems are serious blocks to re¬
liable research, but they are, perhaps, less important where learner training is
concerned.

Factors affecting strategy choice

Learners have been found to vary considerably in both the overall frequency
with which they employ strategies and also the particular types of strategies
they use (O’Malley et al. 1985a; Chamot et al. 1987 and 1988; Ehrman
1990). A key question is whether this variation is systematically related to L2
proficiency and, more especially, to differential success in learning an L2—a
question we take up in the next section. Here we consider the range of factors
that have been found to affect strategy choice.

Individual learner differences

Chapter 11 considered individual learner differences in terms of (1) attitudes,

(2) affective states, and (3) general factors. There has been little research
 Learning strategies 541

into the effect that learners’ affective states have on strategy choice, so we will
focus here on how attitudes and general factors influence learners’ use of
learning strategies. We will also consider how learners’ personal back¬
grounds affect strategy use.

Beliefs about language learning

Bialystok (1981b) found that Grade 10 and 12 learners of L2 French in
Canada varied in the extent to which they believed that language learning in¬
volved formal as opposed to functional practice, and that this influenced their
choice of strategies. Wenden (1987a) also found that learners who emphas¬
ized the importance of learning tended to use cognitive strategies that helped
them to understand and remember specific items of language, while learners
who emphasized the importance of using language employed few learning
strategies, relying instead on communication strategies. Learners who
stressed personal factors did not manifest any distinct pattern of strategy use.

Learner factors
Age emerges as a clear factor affecting the way strategies are used. Young
children have been observed to employ strategies in a task-specific manner,
while older children and adults make use of generalized strategies, which they
employ more flexibly (see Brown et al. 1983, cited in O’Malley and Chamot
1990). Young children’s strategies are often simple, while maturer learners’
strategies are more complex and sophisticated. For example, Brown et al.
found that ‘rehearsal’ for children consisted of rote repetition, while for ad¬
ults it involved ‘active, systematic and elaborative procedures’. Ehrman and
Oxford (1989) also report adults using more sophisticated strategies. These
differences may help to explain why older children and adults generally learn
faster initially than young children (see Chapter 11, page 485) and also why
this advantage is more evident in grammar and vocabulary, for which there
are many learning strategies, rather than pronunciation, for which there are
few. However, as we saw in Chapter 11, there are other explanations for age
differences in rate of learning.
Aptitude does not appear to be strongly related to strategy use. Bialystok
(1981b) found that aptitude was not as influential as learners’ beliefs. How¬
ever, it is possible that learners with enhanced decontextualized language
skills (seen by Skehan (1989) as one aspect of aptitude) will be better able to
talk about the strategies they use. Leino (1982) found that learners with high
conceptual levels were better at describing their strategies than learners with
low conceptual levels. It is possible, then, that learning strategies are related
to that part of language aptitude shared with a general intelligence factor.
Oxford (1989) claims that ‘it is likely that a strong relationship exists be¬
tween the individual’s use of learning strategies and the individual’s learning
style’, but admits that little research has examined the relationship. The
542 Explaining individual differences

descriptors that Willing (1987) provides of the four learning styles he iden¬
tified in ESL learners in Australia (see Table 11.6, page 507) are suggestive of
the different learning strategies that might be associated with each style.
The strength of learners’ motivation can be expected to have a causal effect
on the quantity of learning strategies they employ. Oxford and Nyikos
(1989), in a study of students of foreign languages in universities in the United
States, found that ‘the degree of expressed motivation was the single most
powerful influence on the choice of language learning strategies’ (1989: 294).
Highly motivated learners used more strategies relating to formal practice,
functional practice, general study, and conversation/input elicitation than
poorly motivated learners. The type of motivation may also influence strategy
choice. In the same study, Oxford and Nyikos reported that formal practice
and general study strategies were more popular than functional practice strat¬
egies, perhaps reflecting the students’ strong instrumental goal of fulfilling
course requirements and obtaining good grades in a programme that stressed
analytical skills. However, in a different context an instrumental motivation
can result in a preference for more communication-orientated strategies, as
Ehrman’s (1990) study of adult students at the US Foreign Service Institute
learning languages for career reasons showed. Politzer and McGroarty
(1985), in a study we will consider in detail in the next section, also make the
point that learners’ goals are likely to determine strategy use. For example,
the strategy of asking the teacher how an expression is used may be seen as
relevant by the student concerned with developing communicative compet¬
ence, but not by one concerned only with reading technical literature in the
L2. It does not follow, of course, that students will always be correct in their
assumptions about which learning strategies are best suited to their particular
goals (see Abraham and Vann’s (1987) account of Pedro’s inflexible use of
learning strategies).
It is intuitively appealing to hypothesize a close relationship between per¬
sonality types and strategy choice. Erhman (1990) used the Myers-Briggs
Type Indicator to measure overall personality type with teachers and students
at the Foreign Service Institute. The general approach Ehrman takes is to sug¬
gest that each personality trait is associated with ‘assets’ and ‘liabilities’
where language learning is concerned. For example, extroverts are credited
with a willingness to take risks (an asset) but with dependency on outside
stimulation and interaction (a liability). However, she offers little evidence in
support of such claims. The relationships between individual traits and re¬
ported strategy use were also puzzling in some cases. It is to be expected, per¬
haps, that extroverts should report greater use of affective and visualization
strategies, but other relationships noted by Ehrman are counter-intuitive. For
example, another finding was that introverts reported significantly greater
use of strategies that involved searching for and communicating meaning
than did extroverts, which is surprising given that extroverts have been found
to be generally more successful at acquiring basic interpersonal communica-
 Learning strategies 543

tion skills than introverts (see Chapter 11). Several of the other results re¬
ported by Ehrman and Oxford are also quixotic—for example,
‘feeling-people’ reported using general study strategies to a greater extent
than ‘thinkers’. Such results suggest that the measurements of personality
and/or strategy use may not be reliable. If there are important links between
personality and strategy choice, they remain to be demonstrated.

The learner’s personal background

There is considerable evidence to support a link between learners’ personal

backgrounds and strategy use. Ehrman (1990) found that professional lin¬
guists reported using more strategies more frequently than untrained in¬
structors and students. Also, students with at least five years of study reported
using more functional practice strategies than students with four years or
fewer. Chamot et al. (1987) also found that higher-level high school students
of Spanish and Russian in the US reported using more strategies than did be¬
ginning-level students. However, O’Malley et al. (1985a), in a study of ESL
high school students, found the opposite, although this may have been due to
the fact that the interviews with the beginners were conducted in their mother
tongues while those with the more advanced learners were carried out in the
L2. The general superiority of more experienced language learners over less
experienced is again evident in one of the few longitudinal studies of learning
strategies. Chamot et al. (1988) found that novice high school learners of a FL
were likely to panic when they realized they lacked procedural skills for solv¬
ing a language problem, whereas expert learners (defined as those who had
studied another FL previously) approached tasks calmly and were able to em¬
ploy the strategies they had developed elsewhere. Nation and McLaughlin
(1986) also provide evidence of the superiority of experienced language learn¬
ers over inexperienced. They taught groups of monolingual, bilingual, and
multilingual subjects an artificial language and found that the multilinguals
did better on an implicit learning task, a result they explained by suggesting
that multilinguals were more able to utilize learning strategies automatically.

Situational and social factors

Individual learner differences constitute one source of variation in the use of
learning strategies. Another source is situational factors: the language being
learnt, the setting in which learning takes place, and the tasks that the learner
is asked to perform. Factors of a social nature, such as sex, have also been
shown to influence strategy use (see Figure 12.1, page 530).
There is evidence to suggest that learning some languages results in greater
strategy use than learning others. Chamot et al. (1987), in the study referred
to above, found that FL students of Russian reported greater strategy use than
students of Spanish. Politzer (1983) found that students of Spanish used
544 Explaining individual differences

fewer strategies than those of French and German. However, as Oxford

(1989) points out, this may be because it is more able students who elect to
study the languages less commonly taught in schools in the US.
A number of differences between the learning strategies used by learners in
a classroom as opposed to those used in a more natural setting have been
found. Studies of classroom learners suggest that social strategies are rare.
Chamot et al. (1988), for example, noted that their classroom learners men¬
tioned social and affective strategies infrequently, the only exception being
‘questioning for clarification’. They suggest that the adult-student interview
situation may have inhibited the occurrence of strategies such as ‘co¬
operation’ and ‘self-talk’. It is likely, however, that in many classrooms the
kind of interaction that takes place affords little opportunity for the use of
social strategies (see Chapter 13). Wong-Fillmore (1976; 1979), however,
reports extensive use of social strategies by young learners in a play situation.
There may also be differences in strategy use according to whether the
classroom setting is a second or foreign language one. The FL students invest¬
igated by Chamot et al. (1987) claimed to use some strategies not mentioned
by O’Malley et al.’s (1985a) ESL students (for example, rehearsal, transla¬
tion, note-taking, substitution, and contextualization). The FL students also
reported relying on cognitive strategies (in relation to metacognitive and
socio-affective strategies) to a lesser extent than the ESL students.
It is likely, though, that it is not so much macro-differences (such as the
FL/SL distinction) as micro-differences to do with the specific learning set¬
tings in classrooms that have the greater effect on strategy use. Evidence for
this comes from the effect that task type has been found to have on strategy
choice in the series of studies conducted by O’Malley and Chamot and their
associates. O’Malley et al. (1985a) found that ESL learners reported using the
highest frequency of strategies for vocabulary learning tasks and oral drills.
The lowest frequencies were for listening comprehension, inferencing, mak¬
ing an oral presentation, and engaging in operational communication. Cha¬
mot etal. (1987) and Chamot et al. (1988) provide evidence to show that task
type had a marked influence on learners’ choice of both cognitive and meta¬
cognitive strategies. For example, vocabulary tasks led to the use of the cog¬
nitive strategies of ‘resourcing’ and ‘elaboration’ and the metacognitive
strategies of ‘self-monitoring’ and ‘self-evaluation’, while listening tasks led
to ‘note-taking’, ‘elaboration’, ‘inferencing’, and ‘summarizing’ as cognitive
strategies, and to ‘selective attention’, ‘self-monitoring’, and ‘problem-
identification’ as metacognitive strategies.
These studies suggest that learners’ ability to use a broad range of strategies
flexibly may depend crucially on the nature and range of the instructional
tasks that they experience in the classroom. However, it may not always be
possible to predict the kinds of strategies that an individual learner will
employ in performing a particular task. Hosenfeld (1979) showed that
one learner adapted a language drill to her own preferred approach by
 Learning strategies 545

immediately seeking to apply the grammatical explanations provided by the

task to an imagined communicative situation. Specific tasks may predispose
learners to use particular strategies, but they cannot predetermine the actual
strategies that will be used.
Less attention has been paid to the role of social factors such as
socio-economic group, sex, and ethnicity on the use of learning strategies.
However, Oxford and Nyikos’ (1989) study found that sex had £a profound
choice on strategy choice’ in their study of university students learning
foreign languages. The results tend to bear out previous studies of male-
female language differences (see Chapter 6). Thus, females used conversation
input elicitation strategies more frequently than males, perhaps because they
were more orientated towards social interaction. Ehrman (1990), in a study
of teachers and students in the Foreign Services Institute, found that females
reported greater overall use of strategies than males.

Factors in strategy use

To sum up, there is now evidence to suggest that a number of individual learn¬
er differences and situational factors are related to strategy use. Learners’
beliefs about language learning, the learner factors of age and motivation,
and personal background emerge as important factors. So far, however, there
is only weak evidence to connect the learners’ affective states, language apti¬
tude, learning style, and personality to learning strategy use. In some cases it
is possible to suggest what factors might contribute to superior strategy use
(for example, older learners who are strongly motivated and who have previ¬
ous experience of language learning are likely to use more strategies more fre¬
quently). In other cases, the factors result in different but not necessarily less
successful strategic approaches to learning. Situational and social factors that
have been found to be important are: the language being learnt, the learning
setting, the type of learning tasks, and the learner’s sex. The fact that learning
strategies vary according to learning task suggests that it might be possible to
change learners’ strategic behaviour through training. Such intervention will
only be viable, however, if information is available about the contribution
that different strategies make to learning outcomes.

Learning strategies and language learning

Most of the studies that have investigated the relationship between learning
strategies and language learning have been cross-sectional and correlational
in nature. In one set of studies, the strategies that ‘good language learners’ re¬
ported using have been examined. In another set of studies, the relationships
between strategy use (with regard to both frequency and type) and criterion
measures of learning have been explored using statistical techniques. There
have also been a number of interesting studies that focus on the strategies
546 Explaining individual differences

used to learn vocabulary. The majority of studies have examined adult L2

learners but there are also a few studies of the learning strategies used by
children.

The ‘good language learner’ studies

The ‘good language learner’ studies include most of the early studies of learn¬
ing strategies already referred to, together with some more recent ones (see
Table 12.3). Two approaches have been followed. In one, successful learners
are identified and interviewed and/or asked to complete a written question¬
naire (for example, Naiman et al. 1978; Tennon 1989). In the other, compar¬
isons of more and less successful learners are made (for example, Reiss 1985;
Huang and Van Naersson 1985).
It is easy to overstate the commonalities in strategy use among good lan¬
guage learners. Stevick (1989), in a study of seven successful language learn¬
ers, noted that they ‘differ markedly with regard to what... they prefer to do
and not to do’ (1989: 128). He nevertheless thinks that it is possible to identi¬
fy an ‘overall pattern’.1 This is what we will try to do.
There are, perhaps, five major aspects of successful language learning, as
evidenced by the various studies summarized in Table 12.3: (1) a concern for
language form, (2) a concern for communication (functional practice), (3) an
active task approach, (4) an awareness of the learning process, and (5) a capa¬
city to use strategies flexibly in accordance with task requirements.
Rubin’s (1975) study listed ‘attention to form’ and ‘monitoring one’s own
and other’s speech’ as key strategies. Naiman et al. (1978) include ‘self¬
monitoring and critical sensitivity to language’ as an important character¬
istic. They found that good language learners treat language as a system by
making effective crosslingual comparisons, analysing the target language,
and using reference books. Such learners also monitor their L2 performance
and try to learn from their errors by asking for corrections when they think
these are needed. Reiss (1985) found that ‘monitoring’ and ‘attending to
forpi’ came out as the most common strategies used by learners whom teach¬
ers picked out as ‘good’. Reiss comments that ‘it is surprising that attending to
meaning was of less importance than attending to form’. However, Huang
and Van Naersson’s (1985) study found no difference between high- and low-
proficiency groups on two strategies that reflected attention to form (‘formal
practice’ and ‘monitoring’), although this may have reflected the fact that
learners were asked to report on the strategies they used outside the class¬
room. Gillette (1987) also reports that her two successful learners did not pay
much attention to learning conscious rules, although she notes that they used
their errors as a tool for learning, suggesting, perhaps, that they did make ef¬
forts to attend to form. Overall, there is convincing evidence from the good
language learner studies to show that paying attention to the formal
 Learning strategies 547

cr
0
0
*o
c
0 0 O
C T-*
0 c

Successful learners also gave more specific answers to

0) o 0
o
O O 0 o
0
o X g O
c
(0 0
‘c 0
o o
CD c 3 a
o -C
03 o E 0 03
03 o 0
0 o E o
C/5
c o O c
-X
0
o 03 o E
E c
o o 03
c o
0
0
0 x: "U
o 0
C E
0
0 0
-O
CD
I
o el
0
■o c
0 '
>
0 c ■O ■o
Q) O) TO
0 c 0 CO 0
0 0 0 0 03 o „ c .£ m
03 Q_ 0 ■g to •— t- ©
o 0 n> 0 E E c 0
■O 0 -C o 0 0 0 o 0 ® E TO >
0
0
0 03 0
.E q3 03
■G
0
si 0
~o
C
0
c
0
0 — 000
O 0 E 0
0
3 > ra ®
"O
— c
0 i_ E
0 0 0 -£ .0, 03 0 0
0 O
0 i 0
0
O
.<£> 0 O ■E o 03
03
0 -C 0
03
0 o E tr "(f)
0 0 -r; c
03
CL 8 © °' -C

ID - c 0
0
c c -o
0 2 3 3 O
© w 3 3 2 0 0 c 0
3 c 0 03 03 $=
0 H
O' 0 .9 > Sc|
0
0 O
05
E
CL
CL C ^0
0 0 2 C C 0 'c 0 ^
g E ; -E o
O :«ic
E 0 E q> £ 0. c 0 0 o -S5 0 0 2 <0
S§ -2 o ^—
o O o E i & O Q Q
S <fl
0 0 o 0 O 0 o o
0 0 O O c ^ 2 0 2
^ 0 ll 2 ai o s.® °

questionnaire
C v -2 I
0^
'Co)4- C ^ 03 to to E .2
*3 C o
0 "O " © P C 03
.E o c
0 0
0) 0
0 0
^0^0
o E c .
0
0 -*
3 03
o 2?
E .2
0 0
©
.Pi
- ©
0
a) c 0 ■Jji
2 o 0 ©
032? Q_ CO o "O
E! N N
O 03 O 0 ~ Q. 0 g So ^ o .9
_© 0 O 0 E
C 0 0 0 •- r
E c c 03 E c E
3
> CL 0 — 0 o
0 o “ F i 11 ■0 "0
0 0
* c c E
0 o 0
>
© Jr o 0 i s P ^ ^ 0^ v2
© £ £ £ E
0
0) ll
o
*2 ^ £
0^0 LeI
?<
CC CC .E ^ 2 o
0 1 J to
-O Q.
(f) t— cm co
E E
m co O t- cm co
CC LLt— ll in id 0 < C

03
_ c
3 0
o 0
0
0 0 IS
0 it=
0 03
0
03
c
o 0
c ~o O O C) 0
0 c o o 0 ~o
11 0 3 3 CL 0 0
0 0
o ■o 0 0 0
O
CL 0 c
O ■0
E O c 0 O g
0 0
'■5 1 0 CL 0
_c
3 o
0
0 E ■o •D E CL
N 0 0 >-
0
0
0 -C
o -I x:
CO
x: 0 -c
CL
E
I
'0
c
o
o 03
a §
03

03
C 0 0 L
Is C
o £
>
0
0
©
0
0

D E 0
0
0
n o
M—
~o
| £
0
Q.
CC 0 ® -c ■§ o c 0 0 0
II °
o 5-Is
g- TO °
“
E.2 S o -Q cn
c 0 O a £ 0
O 0 Q ■g T3
O ■§
0 >
?> ~o °
0 0 o 2 £
03
c 3 O 0
CL -£
iO go
■D £ o © 0 0 C
© S
O £ 0 W 0 Jrw 0
© 2 0 0 X) 0 c
x: 0 03
© 3
0 sg
0
0 C .E "o o O O 3 X3

5 S'l T- CM CO O '0

.c O
O 0
O CM
0
03
C r-
2-0
0
0
0
s
0 0
"O 03
0 C
X 0
c
0
E
LL C
CM ©
o
c 111 c Ep
0 © O ^
CO 03 © CO
° E ■o C c 0 TD loo®
0 0 0 (/)
2 o 0
■O
0 O
E ©
‘ >E:
0 0
03
c g IP |
CO 0 03 D t:

—1 o
0
0 0
.0 t- CM
o
>- 0 ml

CO
in CO CO
h- 03 03
03
i—
0
> c 0
~o J0 '0
3 3 © 03
4-» CC CC
</) CC
548 Explaining individual differences

_C
c o
o
c
o
0 0
0 -* .g>
0
o 0 ^
0 0 r2i]
03 .* “ c "O --
O Q.
O
?! cn a
-O .0 C 03 0 »|£S
‘r> 3) '
2 C 0 t0 S °o c i: z .2 -o
0 0 0 C 0
® s O -C
2- 8 •“ O
Q-‘>- 0
Q. "POJfeN E
i|B "O
03
CL 0 C
0 o t §B
o ■
o c o
— O 03 C _ § * | § ^ 'll) TJ •;
0 c > ZF
E e a o o E oE o£|S
0 C 2 «
0 0

n0 _0
— i ° ro « ?
« ? To 0
-o 03 o 03 .i. >>0 > — B Q. ® o
O) 0) -— C Z ^ 0 0 4-j
zoo LU 0 ? S3 8
o >
0
' '—
3,^‘i
0
m r ° o
” O o
i-
ii1 03 0C 7=C o _q c
0 0 F~ C
C 0 ro 0 .2u ft)03
E”
E
5 - E Zc ■p0
8 J3 o=S
01 o m- 0 ^ E 2 0 « 5); 5
000
n E w Z O
0 0 0 0 0 3' 5.-®
03
00 O O ,n c0
E c c 2 »
« 5 -I « w- 1° liif
:> 5 -
0 E. .2
8 ° 0
0 f®> 0)0,
03
C
“O .E 0
O
II S-y ® ra
2 8 c-o &-2
0 tfc o 0 T3 c o E Q_ o O 0 -O ^
■I S
.
5 "d
CO -w
®
>- tC
®
£ 0
CNj Q.
0 0
0
03 0
Z _0 0
E E 03 ~ o 2 0 0 |! 0 CL O E 2
0 ~ 0 o
o o 0 03.9
E c
8I?1 03 2
o
c
0^ «80 « ° it Is
0 O) §
.z
03 0 3 .p
~ 0 03 ^ ~
O’
■ "O 0

Studies of the good language learner

i~
f- F 0i- 0o C 03 o 0 0 C a o ~ .2 0
3 O
± _0) 0 ip
®°®.5 'co .Ec 00 'o E "o E
E (/)
0 C ‘r-
' i- tz.
O 0 03
0 2
0 o
03 'F C
2 o £ er ZEo 0Eo 4-}=;°= -5i=0 o00 0 2 0 -C
2 -8 ^ I 2
03<D 0Z^(/)
Q- f—
rn t- J E I 0 O 2 | ® E £
er
-C
f— •»- CNJ CO CNJ
03 0
co a:
ll » to 3 .9 2
O 0 — 0 -Q Q. Q. 0 <
> ^ E o in
0 0 O "O

0
0 E £
-Q B E
0
"i_ 0 -Q
0 2 0
0 > 03 0
Q. >• 0 O
Z 0
0
c
E 03 C
o 03 0
O c 0 r- 0
0
bI CL
?§ 0 C4-
0
C
s § 3 ■4= 0 ^ o
C
0 JZ o
=> E O 0 0
-Q 0 0
0 "O JD 0
0 0
§?
0 0 0 0
h:
.h:
0
0
'
O
O
0
3 c
0
C O’
0 0~ > 0 03 C ■0 c
03
- £ o 0 O Z
o 0 S ^
EB I
Taitfe 22.3;

C 0 0 2 0
E 0 -2
■2 § '2
o z 0 0 0 O 0
4= 0
§ c o > E 0 ^
o
S 0 03
3 .E
-77 Co o
<;
0
2
C
O’ E 0 Z E E C ~
o 0 0
"rr*
03 0 o
TJ
C O Q. 2 S •D C | | c c
o
o
.c IIS 0
c
—
-O 0 z
C 0 o -o
0 0
® 2
0
Z 0
4-* 0 c z 03
0) g2i 03 0 0 ^ 0 C5 . ^
0
CL
2 T- C\J y- CNJ

0C 0
E
Z
O o 03 Ccj O
"8
0
c
0 I 0 co ®
>
O< <f)0 JZ
^
•-
0
_0 3gl
0 "O
0
o
o
0 0 U_
If
sEl 03
°§ “
“O ^ LU E a)
® J.
c
0 O O S
— iS "o ■o
o m *t=;
® .5
03 -C 0
2 ® S2 0 C 0 0 0 0 E x:
n
■p
03 o 0
f ? E ® r
Z E c E 0
° 2 0 O 0 2 0 0 U o)
Q. O
CNJ Q. -O O 0 .E "O 0 CO .E LU

E
> 03 N- 03
-o 1” If) CO 00
C C co 03 03
0 O 03 i— T—
03 ^ 0 C
> 0 o
c 2 z c
0 0 0 0
3 0 0 — c
CO I z 0
oc b _l
 Learning strategies 549

properties of the target language contributes to success—a view that seems

to conflict with the claims of Krashen’s (1985) Input Hypothesis (see
Chapter 7).
Good language learners also attend to meaning. All the studies listed in
Table 12.3 refer to the importance of this aspect of strategy use. Indeed,
Huang and Van Naersson found that ‘functional practice’ distinguished the
more from the less successful learners in their study, whereas ‘formal practice’
and ‘monitoring’ did not. In most of the studies, however, the learners ap¬
peared to benefit from attending to both form and meaning. For example,
Gerardo, the more successful learner in Abraham and Vann’s (1987) study,
paid attention to both, while Pedro, the less successful, paid attention only to
meaning. Good language learners search for meaning in the L2 data they are
exposed to and try to engage in real communication by seeking out opportun¬
ities for natural language use. Naiman et al. (1978) reported that the learners
they studied emphasized fluency in the early stages, giving more attention to
form later. Lennon’s (1989) study of very advanced learners of L2 English
showed that they were adept at alternating between a focus on learning the
language and on communicating in it. The ability to switch to and fro in at¬
tending to meaning and form may be a crucial feature of successful language
learning.
Good language learners show active involvement in language learning.
They appreciate teachers who are systematic, logical, and clear, but prefer to
treat them as ‘informants’ rather than to rely on them (Pickett 1978). They
like to take charge of their own learning by identifying and pursuing goals
and by trying to introduce new topics into a conversation. However, being
‘active’ does not necessarily mean engaging in language production. Reiss
(1985) reported that many successful classroom learners were ‘silent speak¬
ers’ (i.e. they rehearsed and practised silently while listening to others)—a
finding that will be picked up in Chapter 13 when considering a number of
studies that indicate listening may be as important, if not more so, than oral
participation in classroom language learning.
The fourth general characteristic of the good language learner—awareness
of the learning process—suggests the importance of what O’Malley and Cha-
mot have called ‘metalingual strategies’. Successful learners are thoughtful
and aware of themselves in relation to the learning process. They take con¬
scious decisions and they follow their own preferred learning style. These are
the learners who are able to talk effectively about their language learning be¬
cause they have a well-developed metalanguage for doing so. Reiss (1983)
found that students awarded ‘A’ grades by their teachers were able to give
very specific accounts of how they would approach different learning tasks,
while learners with lower grades were often vague and imprecise. For ex¬
ample, asked to comment on how they would set about learning a new task
that had been explained in class, the ‘A’ students gave answers such as the
following:
550 Explaining individual differences

‘Try to practise the new tense while speaking.

Look for similar endings to those already known.’

while ‘B’ students gave answers such as this:

‘Keep going over it.

Study it until I understand.’

Good language learners also make use of metacognitive knowledge to help

them assess their needs, evaluate progress, and give direction to their learn¬
ing. Such awareness gives learners control over their own learning.
One of the most comprehensive studies of learner strategies to date,
Chamot et al. (198 8), provides evidence in support of the final general charac¬
teristic of good language learners—flexible and appropriate use of learning
strategies. Chamot et al. investigated beginner, intermediate-, and advanced-
level students of Spanish and Russian over four school semesters. The stu¬
dents were classified as ‘effective’ or ‘ineffective’ by their teachers. No clear
pattern was evident in the strategies the learners used from one time to the
next, although changes were evident in the performance of specific tasks over
time. For example, the beginners more than doubled their use of strategies in
a writing task and also showed a number of more qualitative changes in strat¬
egy use (for example, greater use of ‘comprehension monitoring’). However,
what set the ‘effective’ students apart was their use of a greater range of strate¬
gies and, in particular, their ability to choose strategies that were appropriate
for particular tasks. The effective learners were also more purposeful in their
approach, engaged in ‘comprehension monitoring’ to a greater extent than
‘production monitoring’ (involving attending to separate linguistic compon¬
ents), and made extensive use of their general knowledge as well as L2 lin¬
guistic knowledge.
Studying good language learners has proved a useful way of investigating
how strategies affect language learning. It has afforded both generalized com¬
ment and fine illustrative detail. It is not clear, however, how much store
should be set by the studies undertaken to date. They have focused mainly on
classroom learners and the ‘good’ strategies that have been identified neces¬
sarily reflect the formal learning setting. The main methods of collecting
data—learners’ verbal reports—may give an advantage to the kind of learner
who is able to talk about language learning easily and skilfully. This may be
why ‘metacognitive’ strategies have been so strongly emphasized. Also, it is
not clear whether the strategies that have been identified are the cause or the
result of success. Despite these reservations, the good language learner studies
have provided some of the richest insights into the kinds of behaviours associ¬
ated with successful language learning. They constitute one of the most effect¬
ive lines of enquiry in learning strategy research.
 Learning strategies 551

Correlational studies employing statistical procedures

A number of studies have sought to examine whether there are specific strat¬
egies that are statistically related to L2 proficiency. The results obtained by
these studies are not always clear, but they lend some support to the main
findings of the ‘good language learner’ studies.
In a series of studies, Bialystok explored the relationship between four
strategies and L2 proficiency. Two concerned the learning resulting from
communicating in an L2 (‘functional practice’ and ‘inferencing’), and two
concerned more conscious attempts to learn the L2 (‘formal practice’ and
‘monitoring’). These studies were informed by Bialystok’s Model of Second
Language Learning, which was described in Chapter 9. In her main study,
Bialystok (1981b) investigated students studying French in Grades 10 and 12
in Canada. She used a questionnaire to collect information on the students’
reported use of the four strategies in both oral and written tasks. Proficiency
was measured by means of oral and written tasks that required attention to
meaning and to form. The results are not easy to interpret and in some re¬
spects are counter-intuitive. For example, Bialystok failed to find any rela¬
tionship between inferencing and proficiency, although elsewhere a strong
theoretical case for the importance of inferencing in L2 learning has been
made out (see Carton 1971; Bialystok 1983a). Another result that seems a
little odd is that only functional practice correlated significantly with profi¬
ciency in the Grade 10 students, whereas three strategies (functional practice,
formal practice, and monitoring) were related to proficiency in the Grade 12.
It is possible, of course, that this reflects a shift in strategy use as proficiency
develops, but it is also possible that the data elicited by means of the question¬
naire were not reliable.
If Bialystok’s attempt to show a statistical relationship between strategy
use and proficiency was only partly successful, Politzer and McGroarty’s
(1985) attempt was even less successful. This study focused on a much
broader range of strategies, but once again made use of a questionnaire. Al¬
though considerable attempts were made to ensure the validity of this instru¬
ment by basing its contents on previous studies of good language learners, the
generally unsatisfactory nature of the results obtained again suggests that ‘re¬
ported strategy use’ may not provide an accurate account of what learners ac¬
tually do, a problem that Politzer and McGroarty acknowledge.
The study elicited information about the behaviours learners reported
using in (1) study inside the classroom, (2) individual study, and (3) social in¬
teraction outside the classroom. The subjects in this study were 37 learners
from two major ethnic groups (Flispanics and Asians) enrolled on an eight-
week intensive ESL course at American universities. Learning gains were cal¬
culated from scores on three tests (a comprehension test, a discrete item test of
linguistic competence, and a test of communicative ability), which were ad¬
ministered at the beginning and the end of the course. Very few statistically
552 Explaining individual differences

significant correlations were found. Social interaction strategies were related

to gains in communicative ability, but no other relationships between the
three general groups of strategies and the criterion measures were found.
However, different clusters of strategies were found to be related to gains in
different tests. For example, reported behaviours associated with active in¬
quiry concerning language use (for example, ‘asking teacher about an expres¬
sion’ and ‘asking for confirmation of correctness’) were correlated with gains
in listening comprehension and communicative ability, while reported beha¬
viours involving attention to form (such as ‘keeping track of new vocabulary’
and ‘trying to use new words’) were linked to gains in linguistic competence.
Politzer and McGroarty argue that this was the most interesting finding. It
points to a general weakness in- much of the learning strategy research,
namely that strategies have tended to be considered in isolation rather than in
groups.
The final correlational study we will consider is in many ways the most in¬
teresting and the most insightful. Mangubhai (1991) studied the strategies
used by five adult beginner learners of L2 Hindi who received four weeks of
instruction by means of Total Physical Response (Asher 1977)—a method
based mainly on the use of oral commands that students have to listen to and
carry out. Information about the learning strategies they used came from con¬
current think-aloud tasks, immediate retrospective reports, and discussions
at the end of each instructional session. Three sets of strategies were iden¬
tified, depending on whether the focus of the different behaviours was (1) on
form, (2) on meaning, and (3) on memory (retrieval and storage). Achieve¬
ment was measured by means of a test consisting of oral commands similar to
those used in the instruction, a sentence repetition test and a listening compre¬
hension test. On the basis of these tests, the learners were divided into two
groups: Group A, consisting of three ‘high’ achievers and Group B of two
‘low’ achievers. Table 12.4 summarizes the main differences in strategy use in
the two groups of learners. This shows that the ‘high’ achievers used more
memory strategies, were more likely to direct their attention to chunks than
to individual words, relied less on translation, and paid more attention to the
form of the commands once they had extracted the meaning. The strength of
Mangubhai’s study is that it elicited information relating to strategy use in a
variety of ways, making it possible to compare what learners reported doing
retrospectively with what they reported doing during an actual language
learning task.
These three studies—all of which investigated classroom language learn¬
ers—afford only limited information about the relationship between learning
strategies and L2 learning. Two of them (Bialystok’s and Politzer and
McGroarty’s) have produced rather indeterminate results, perhaps because
they relied on learners’ reported use of strategies in a written questionnaire.
Mangubhai’s study overcame this limitation and also benefited from in¬
vestigating complete beginners over a period of time, thus allowing us to
 Learning strategies 553

Group A Group B

1 Lesser focus on meaning relative to 1 Greater focus on meaning relative to

memory memory
2 Greater focus on memory relative to 2 Lesser focus on memory relative to
meaning meaning
3 Lesser focus on single words under 3 Greater focus on single words under
meaning meaning
4 Less translation into English 4 More translation into English
5 More practising 5 Less practising

Table 12.4: Differences in strategy use between high and low achievers in a four-
week Hindi course (Mangubhai 1991:282)

tentatively attribute a causal role to the strategies his learners employed. The
results suggest that a learning approach that might be described as ‘audio-
lingual’ in nature works well,2 but this may simply reflect the ability of the
successful learners to adapt to the particular learning task they were asked to
perform. A study of five learners, of course, does not permit generalization.
These three studies also investigated different sets of strategies, reflecting one
of the main problems of strategy research, namely the lack of agreement re¬
garding what strategies and what combinations of strategies should be stud¬
ied. It is not possible to reach any general conclusion on the basis of them.

Studies of vocabulary-learning strategies

A number of specific strategies for learning vocabulary have been identified
(see Nation 1990; Cohen 1990; Oxford 1990). A general distinction needs to
be drawn between those strategies used to memorize isolated lexical items
and those strategies used to learn new words from context. We will consider
two studies, one longitudinal and the other experimental.
A study by Cohen and Aphek (1981) suggests that the use of mnemonic as¬
sociation aids vocabulary learning. It investigated seventeen English-speak¬
ing students (nine beginners, six intermediate, and two advanced) learning
Hebrew over a 100-day period. Not surprisingly perhaps, the learners’ main
approach was simply to try to memorize the words they did not know. Most
of the learners reported using associations of various kinds and Cohen and
Aphek were able to identify eleven different types, involving the target lan¬
guage, the learners’ LI, and extralinguistic signs. The general conclusion was
that any attempt to form an association involving the target word aided reten¬
tion. In contrast, the frequency with which the learners reported encoun¬
tering the items outside the classroom had no effect on retention. Again,
though, we should not note that what learners report experiencing and what
they actually experience might not be the same. Another interesting finding
of Cohen and Aphek’s study was that there was an interaction between the
554 Explaining individual differences

learners’ overall level of proficiency and the kind of task that worked best for
vocabulary learning. Beginners found tasks that involved listing best, while
the intermediate learners found tasks that involved contextualization more
effective, suggesting that contextualization strategies work better for learners
who already possess a fair level of L2 knowledge.
A study by Brown and Perry (1991) investigated the success of three vo¬
cabulary-learning strategies: (1) keyword, (2) semantic, and (3) keyword-
semantic. The keyword strategy involves asking learners to form a visual as¬
sociation between the target word and some acoustically similar word they
already know (the keyword). For example, to learn the Japanese word ‘ski’ ( =
‘like’), learners might use the keyword ‘skiing’ and imagine themselves en¬
joying skiing down a mountain slope. A semantic strategy involves some kind
of attempt to integrate the target word into the learner’s existing semantic
systems, for example by identifying how a word relates to other known
words. A keyword-semantic strategy involves a combination of (1) and (2).
Brown and Perry hypothesized that strategies that involved greater ‘depth of
processing’ (see Craik and Lockhart 1972) would result in better retention.
Their prediction was that processing at the ‘shallow’ sensory level of the kind
involved in the keyword strategy aids retention less than processing at the
‘deep’ semantic level. Processing that involves both shallow and deep levels
was hypothesized to be more effective still. An experimental study of six in¬
tact upper-level classes at the English Language Institute of the American Uni¬
versity of Cairo lent support to Brown and Perry’s hypothesis; the class taught
to use the semantic-keyword strategy retained significanctly more words, as
measured by cued-recall and multiple choice tests, than the class taught the
keyword strategy, while the results for the class taught the semantic strategy
were intermediate.
The study of vocabulary-learning strategies is a promising area of enquiry.
This is because it is possible to define the learning targets and strategies very
precisely, and also to investigate strategies that have wide currency in the lit¬
erature. Also, vocabulary-learning strategies lend themselves to experimental
investigation, which O’Malley and Chamot (1990) suggest is now needed to
develop the field of learning strategy research. Most important, perhaps,
Brown and Perry’s study shows that it is possible to locate vocabulary strat¬
egy research within a strong theoretical framework (i.e. depth-of-processing
theory).

Studies of learning strategies used by children

The great majority of learning strategy studies have examined adolescents or
adults. However, one or two studies have also looked at children. Perhaps the
best known is Wong-Fillmore’s (1976; 1979) nine-month study of five Mex¬
ican children learning English in the United States. Wong-Fillmore collected
natural spoken data by pairing each learner with a native-speaking child in a
 Learning strategies 555

play situation. Perhaps as a reflection of the way in which the data were col¬
lected, ‘social strategies’ are emphasized. One such strategy is ‘Join a group
and act as if you understand what is going on, even if you don’t’. Linked to
each social strategy is one or more cognitive strategies. For example, the cog¬
nitive strategy linked to the social strategy described above is ‘Assume what
people are saying is relevant to the situation at hand’ and ‘Guess’.
The strategies described by Wong-Fillmore are those employed success¬
fully by children in the early stages of language learning. A study by Chester¬
field and Chesterfield (1985) examined how children’s learning strategies
change as their knowledge of the L2 develops. The subjects in this study were
fourteen Mexican-American children, eight of whom were studied at three
different points during their first year of schooling. Data were collected
through observations in bilingual classrooms. The strategies used by these
children emerged in a distinct order. The ones to appear first were mainly re¬
ceptive and self-contained (for example, ‘repetition’, ‘memorization’, and
‘use of formulaic expressions’). Later strategies to emerge were those based
on interpersonal interaction (for example, ‘appeal for assistance’ and ‘re¬
quests for clarification’), while later still were those that reflect metacognitive
knowledge (for example, ‘monitoring’). Chesterfield and Chesterfield did not
try to find out which strategies contributed most to language development.
Whereas the study of learning strategies in adults has relied largely on self-
report data, that involving children has made use of observational data. This
is reflected in the kinds of strategies identified (i.e. social and discourse strat¬
egies predominate). Studies of adults have emphasized cognitive and meta¬
cognitive strategies. What is not clear, then, is whether the differences in the
kinds of strategies used by adult and child subjects is a reflection of their age
or of the research methodology.

Relating learning strategies to second language development

Caution must be exercised in drawing conclusions from the research that has
investigated the relationship between learning strategies and L2 develop¬
ment. Studies have varied enormously in the kind of learners studied, in the
procedures used to obtain information about strategy use, and in the ways in
which learning ‘success’ has been assessed. The following conclusions, there¬
fore, are tentative:
1 The strategies that learners elect to use reflect their general stage of L2
development. For example, there is some evidence to suggest that strat¬
egies that relate to the functional use of language and that involve pro¬
cessing chunks of language precede those that involve close attention to
form and single words. Metacognitive strategies are more evident in ad¬
vanced learners.
2 Successful learners appear to use learning strategies more frequently and in
qualitatively different ways than learners who are less successful. For
556 Explaining individual differences

example, successful adult beginners seem more adept at the use of memory
strategies.
3 Successful language learning involves attention to both form and meaning.
Good language learners appear able to switch the focus of their attention
while they are performing a task—even as beginners.
4 Different kinds of learning strategies may contribute to different aspects of
L2 proficiency. Thus, strategies that involve formal practice may contrib¬
ute to the development of linguistic competence, while strategies involving
functional practice aid the development of communicative competence.
5 Learners need to employ strategies flexibly by selecting those strategies that
are appropriate for performing a particular learning task.
6 Because of (5), metacognitive strategies involving goal identification,
planning, monitoring, and evaluation assume considerable importance, at
least for adults. However, many learners appear to under-utilize this type
of strategy.
7 The more successful adult learners are better able to talk about the strat¬
egies they use.
8 The learning strategies used by children and adults may differ; social and
interactional strategies may be more important with young learners.

These constitute very general conclusions and it might be argued that not
enough is yet known about the relations between learning strategies and lan¬
guage learning to justify attempts to train learners to use particular strategies.
In particular, not enough is known about what combinations of strategies are
most effective. However, this has not prevented a number of attempts at strat¬
egy training.

Training learners in the use of learning strategies

There is now a wealth of material that has been developed to train learners to
use effective language learning strategies: for example, Ellis and Sinclair
1989; Brown 1989; Oxford 1990; Wenden 1986b and 1991. Somewhat sur¬
prisingly, however, there have been few empirical studies that have attempted
to evaluate the success of this training on L2 learners. We will briefly review
the major studies and then identify a number of key issues in language learn¬
ing strategy training.
One area in which strategy training may be particularly useful is in vocabu¬
lary acquisition. Bialystok (1983b) carried out two experiments to investigate
a number of ways in which the ability of Grade 10 students of L2 French to in¬
ference the meanings of words in a continuous text could be improved. In one
of the experiments, a fifteen-minute lesson on how to inference resulted in
more effective overall comprehension of a written text than providing the
learners with picture cues or letting the learners use a dictionary. However,
dictionary use (but not picture cues) resulted in better scores on a vocabulary
 Learning strategies 557

test than did the strategy training. In the second experiment the strategy train¬
ing proved less effective in promoting either comprehension or vocabulary
acquisition than the other two conditions.
Cohen and Aphek (1980) gave adult learners of L2 Hebrew a short training
session in how to learn vocabulary through associations. The results indic¬
ated that forming associations helped in vocabulary recall tasks and that fail¬
ure to employ an association often led to incorrect recall. Cohen and Aphek
also reported that those students who were more proficient at the outset were
also the most successful in using association in recall tasks, suggesting that
training in forming associations might be most helpful for advanced rather
than beginner learners.
A third study of the effects of strategy training on vocabulary learning
failed to show significant results, but is interesting because it suggests the
complexity of the issues involved. O’Malley et al. (1985b) studied the effects
of two kinds of training on 75 intermediate-level ESL students of mixed eth¬
nic backgrounds (Hispanic and Asian). One group received training in the use
of ‘imagery and grouping’ (a cognitive strategy), while a second group re¬
ceived training in this strategy and also in ‘self-evaluation’ (a metalingual
strategy). There was also a control group. Although no significant differences
among the treatment groups were found, a detailed analysis of the results
showed that while the Hispanic training groups outperformed the Hispanic
control group, the reverse was the case for the Asian groups. Apparently the
Asian students had resisted using the strategies during training, preferring to
rely on rote memorization instead. This study shows that the learning styles
of different cultural groups need to be taken into account in planning strategy
training.3
The study by O’Malley et al. also investigated the effects of strategy train¬
ing on the learners’ performance on a listening and a speaking task. In the case
of listening, the cognitive group was taught ‘note-taking’ and ‘co-operation’
(a social/affective strategy), while the metacognitive group was taught ‘select¬
ive attention’ in addition. In the case of the speaking task, one experimental
group was taught how to use ‘cooperation’ (a social/affective strategy), while
the other was taught both this strategy and ‘functional planning’ (a metacog¬
nitive strategy). In the speaking task, which required students to present a
two-minute talk, the group taught functional planning outperformed the
other experimental group which, in turn, outperformed the control group.
No significant differences were found in the listening task used as a post-test.
Clearly, there are many issues that need sorting out before strategy training
can be implemented effectively. First, more work is needed to discover what
strategies and, in particular, what combinations of strategies should be
taught. Second, ways have to be found of taking into account learners’ own
preferred learning strategies. Third, some learners may need convincing that
strategy training is worth while. Wenden (1987b), for example, found that
students in the American Language program at Columbia University
558 Explaining individual differences

responded negatively to a training component she included in their language

course. As O’Malley and Chamot (1990) point out, the very learners that
need strategy training are most likely to be the ones that reject it. Fourth, it is
not clear whether learner training will work best when it exists as a separate
strand in a language programme (as Wenden tried) or when it is fully integ¬
rated into the language teaching materials, as proposed by O’Malley and
Chamot (1990) in their Cognitive Academic Learning Approach (CALLA). A
fifth issue is when strategy training should be attempted. With beginners it
will only be possible if undertaken in the LI. A final issue is whether learners
should be made conscious of the strategies they are taught, or whether just
providing practice opportunities is sufficient. As Skehan (1991) states, ‘there
is a lot to play for’.

Conclusion
The study of learning strategies holds considerable promise, both for lan¬
guage pedagogy and for explaining individual differences in L2 learning. It is
probably true to say, however, that it is still in its infancy. For this reason, per¬
haps, discussions of learning strategies typically conclude with the problems
that have surfaced and that need to be addressed before progress can be made.
There is no agreement about what constitutes a ‘learning strategy’. There is
no widely accepted theoretical basis for identifying and describing strategies,
although O’Malley and Chamot have anchored their own work, with some
success, in a cognitive theory of information processing, while Brown and
Perry have shown how the study of vocabulary-learning strategies can be in¬
formed by depth-of-processing theory. The work done to date has been essen¬
tially descriptive, reflecting the corpora of data that different researchers have
worked on. As a result, the nature of the strategies that have been identified
varies enormously.
Despite this, considerable progress has been made in developing taxono¬
mies of learning strategies. Where the early research listed, the later research
classifies. O’Malley and Chamot’s three-way distinction between cognitive,
metacognitive, and social/affective strategies is useful and has been generally
accepted. The classification of strategies within these general types remains
more problematic, however.
A lot of the research has been based on the assumption that there are ‘good’
learning strategies. But this is questionable. The beneficial effect of strategies
may be relative to the kinds of tasks they are deployed in. For example, some
strategies may work in tasks aimed at the development of linguistic compet¬
ence and others in tasks with more communicative objectives. Effective strat¬
egy use may consist of the flexible deployment of the right strategies in the
right task, but little is currently known about this. It is also possible that dif¬
ferent strategies are important for classroom and naturalistic language learn-
 Learning strategies 559

ing and for children as opposed to adults. Much of the research to date has
studied the kind of analytic learning found in adult classroom learners.
The general assumption that effective strategy use involves frequent strat¬
egy use is also questionable. It is likely that it is not so much how often learn¬
ers use strategies as when and with what purpose they use them. It is also
likely that strategies will prove most helpful when they are deployed in clus¬
ters, but precisely what groupings work best is not known.
Implicit in much of the research is the assumption that strategies are causal
(see Figure 12.1, page 530). But, as we have seen, a case can also be made out
for treating them as the result of learning. Thus, advanced learners may em¬
ploy certain strategies simply because they are advanced.
Much of the research has relied on learner self-reports, as with the excep¬
tion of some studies of children, most researchers have found observation in¬
effective. Retrospection (through interviews, questionnaires, and diaries) and
introspection (in concurrent think-aloud tasks) have provided rich informa¬
tion on learning strategies, but doubts remain about the reliability of such
methods (see Seliger 1984). Also, there is always the danger that these meth¬
ods conflate the strategies learners use with their ability to self-report.
Finally, nearly all the research has been cross-sectional in nature with the
result that we know little about how learners develop the ability to deploy
learning strategies over time or what effect their deployment has on L2 learn¬
ing. More longitudinal case studies are sorely needed if solutions to the prob¬
lems mentioned above are to be found.

Notes
1 Stevick’s approach to describing the ‘overall pattern’ evident in his study of
the seven successful learners is rather different from that adopted by other
researchers. In fact he specifically avoids using the term ‘strategy’ on the
grounds that this is not used consistently in the literature and is not yet part
of‘the public vocabulary’ (1989:146). Instead, Stevick considers the ‘over¬
all pattern’ in terms of the ‘images’, verbal and non-verbal, and the way
these are locked into memory.
2 I am grateful to Peter Skehan for this observation. He notes that Mangub-
hai’s successful learners were ‘an audiolingual teacher’s dream’.
3 The importance of taking cultural differences into account is also indicated
by Politzer and McGroarty’s (1985) study, where differences between the
Hispanic and Asian subjects’ reported strategy use were noted.

Further reading
There are now a number of full-length books dealing with learning strategies:
A. Wenden and J. Rubin (eds.), Learner Strategies in Language Learning
(Prentice Hall International, 1987).
560 Explaining individual differences

J. O’Malley and A. Chamot, Learning Strategies in Second Language Ac¬

quisition (Cambridge University Press, 1990). This surveys the research to
date and includes useful summaries of their own studies.
R. Oxford, Language Learning Strategies: What Every Teacher Should
Know (Newbury House, 1990). This is a training manual for teachers. It con¬
tains a comprehensive list of strategies and the Strategy Inventory for Lan¬
guage Learning (a valuable diagnostic tool) is made available in an appendix.
For the reader who would like to read some of the original research reports,
a number of the ‘good language learner’ studies are strongly recommended:
N. Naiman et al., The Good language Learner (Ontario Institute for Studies
in Education, 1978). This is sometimes difficult to obtain, but is probably the
single most important study of learning strategies.
M. Reiss, ‘The Good Language Learner: another look’. Canadian Modern
Language Review (1985) 41: 257-66.
A number of other studies are also worth reading. The following serve as
examples of the kind of research that has been undertaken, although the res¬
ults they obtained were not very conclusive.
R. Politzer and M. McGroarty, ‘An exploratory study of learning behaviours
and their relationship to gains in linguistic and communicative competence’.
TESOL Quarterly (1985) 19: 103-23.
F. Mangubhai, ‘The processing behaviours of adult second language learners
and their relationship to second language proficiency’. Applied Linguistics
(1991) 12:268-98.
J. O’Malley et al., ‘Learning strategy applications with students of English as
a second language’. TESOL Quarterly (1985) 19: 285-96.
PART SIX

Classroom second language acquisition

 ■

'

»
Introduction

The previous sections of this book have been concerned with questions of
general import to the study of second language acquisition. They have en¬
deavoured to provide descriptions of how an L2 is acquired, and to explain
the structural processes that account for its universal properties and the social
and personal factors that account for variation in the rate and success of indi¬
vidual learners. Although some attention has been given to the impact that
the acquisitional setting can have on outcomes (see Chapter 6), the primary
concern has been to account for the regularities in acquisition evident across
settings. Thus, no clear distinction has been drawn between the acquisition
that takes place in untutored and tutored settings.
Much of the research that has been reported in the earlier sections, how¬
ever, has involved classroom learners. This is particularly true of experi¬
mental or correlational studies (for example, studies based on the
Competition Model or Gardner’s research on motivation in Canada). One
reason is that the large numbers of learners needed for such research are more
accessible in educational than in naturalistic settings. However, this research
has not been concerned with classroom L2 acquisition per se; rather it has
used classroom learners to investigate questions of general significance to
SLA research.
In this section, we focus our attention specifically on classroom L2 acquisi¬
tion. In so doing, of course, we will revisit many of the general issues raised
earlier in the book—for example, the role of interaction in shaping learning,
the difference between implicit and explicit knowledge, the role of negative
feedback, and the significance of acquisitional orders and sequences. In fact,
the classroom constitutes an ideal setting for examining the key theoretical
issues because it is possible to observe closely how input is made available to
the learner and what kinds of output learners produce in specific classroom
contexts. It is also possible to engineer what input learners are exposed to and
what output they produce in order to investigate specific hypotheses about
how learning takes place.
The reason for focusing on the classroom, however, is not merely to shed
further light on how L2 acquisition takes place. It is also motivated by a desire
to discover what classroom conditions are most likely to facilitate acquisi¬
tion. In other words, it has a pedagogic purpose. However, this pedagogic
purpose is not quite the same as that addressed in methodological handbooks
for teachers, where the aim is to suggest specific techniques or activities that
564 Classroom second language acquisition

teachers can use. The research we will consider in this section invites us to
consider pedagogy not in terms of ‘techniques’ or ‘activities’, but in terms of
what kinds of classroom behaviours teachers need to engage in to promote
learning—what questions to ask, when and how to correct learners’ errors,
how to instigate negotiation for meaning in a classroom, etc. Furthermore,
whereas methodologists base their advice on what might be called ‘sound
pedagogy’, accumulated from centuries of experience of language teaching,
the pedagogical proposals drawn from L2 classroom acquisition research are
informed by theoretical concepts drawn from SLA research. It is not inten¬
ded, however, to suggest that such an approach to language pedagogy is the
preferred one—a point that we will take up in the final chapter of this book.
It is possible to identify two major strands in classroom L2 acquisition re¬
search, corresponding to two ways of viewing the classroom. According to
one view, what happens in the classroom provides opportunities for learning,
which can be explored by examining the relationship between inter¬
action—the means by which the opportunities are provided—and L2
learning. An alternative is to view the classroom as a place where attempts are
made to intervene directly in the process of L2 learning—to ask, in other
words, ‘Do learners learn what they have been taught?’. Chapter 13 examines
research belonging to the first strand, while Chapter 14 looks at the effects of
direct intervention on learning.
13 Classroom interaction and second
language acquisition

Introduction

The classroom affords the L2 researcher three different perspectives. The first
perspective is that found in comparative method, studies. These seek to com¬
pare the effect of different language teaching methods on L2 learning. Such
studies are ‘product’-based because they rely entirely on measurements of
language learning and make no attempt to examine the instructional and
learning ‘processes’ that take place inside the classroom.
A second perspective involves going inside the ‘black box’1 of the class¬
room itself. It views the classroom as a place where interactions of various
kinds take place, affording learners opportunities to acquire the L2. All-
wright (1984: 156) sees interaction as ‘the fundamental fact of classroom
pedagogy’ because ‘everything that happens in the classroom happens
through a process of live person-to-person interaction’. This perspective has
drawn heavily on the research and theories dealing with the relationship be¬
tween input/interaction and L2 learning (see Chapter 7). It leads the re¬
searcher to observe and describe the interactional events that take place in a
classroom in order to understand how learning opportunities are created.
Ideally, it should also lead to attempts to demonstrate the effects of different
types of interactional opportunity on L2 learning, but as we saw in Chapter 7,
few studies have investigated this relationship directly.
The third perspective involves investigating the effects of formal instruc¬
tion. In this case, instruction is viewed as an attempt to intervene directly in
the language learning process by teaching specific properties of the L2. The
question that is asked is ‘Do learners actually learn what they are taught?’ Re¬
searchers have been particularly interested in whether instruction directed at
specific grammatical items and rules has any effect on interlanguage develop¬
ment. This chapter will consider research that has adopted the first two per¬
spectives, and the following chapter will examine research relating to the
third.
First of all, this chapter will consider some of the principal research meth¬
ods that have been used to investigate the role of classroom interaction in lan¬
guage learning. It will then review a number of comparative method studies.
The main part of the chapter is given over to a survey of research that has
566 Classroom second language acquisition

examined different aspects of classroom interaction: teacher talk, error treat¬

ment, teachers’ questions, learner participation, task-based interaction, and
small group work. Finally, consideration is given to the relationship between
classroom interaction and L2 learning and, in particular, whether classroom
environments are capable of providing the kinds of opportunities needed to
develop full L2 competence.
A distinction can be drawn between ‘classroom research’ and ‘classroom-
orientated research’ (see Nunan 1991: 249). The former consists of studies
that have investigated learners inside actual classrooms, while the latter con¬
sists of studies conducted outside the classroom (for example, in some kind of
experimental setting) but which have been motivated by issues of clear relev¬
ance to classroom L2 acquisition. This chapter will consider both types of
research, but will try to make clear which is which.

Methods of researching learning in the second language classroom

A number of different research traditions in L2 classroom research can be
identified. Chaudron (1988) describes four (see Table 13.1). The psycho¬
metric and interaction analysis traditions typically involve ‘quantitative’ and
‘explanatory’ research, while the discourse analysis and ethnographic tradi¬
tions make use of more ‘qualitative’ and ‘descriptive’ methods.

Tradition Typical issues Methods

Psychometric Language gain from different Experimental method—pre- and

methods, materials, treatments.
Interaction analysis Extent to which learner behaviour
is a function of teacher-
determined interaction.
Discourse analysis Analysis of classroom discourse
in linguistic terms.
Ethnographic Obtain insights into the
classroom as a cultural system.
post-tests with experimental and
control groups.
Coding classroom interactions in
terms of various observation
systems and schedules.
Study classroom transcripts and
assign utterances to pre¬
determined categories.
Naturalistic ‘uncontrolled’
observation and description.

Table 13.1: A summary of Chaudron’s four research traditions in L2 classroom

research (taken from Nunan 1990b: 23)

The psychometric tradition

The psychometric tradition is evident in the comparative method studies to be

considered in the next section and in program-product comparisons of the
kind used to evaluate different types of immersion programmes (see Swain
and Lapkin 1982 for a review). The main problem with such studies is that
 Classroom interaction and second language acquisition 567

without a process element to provide information about the actual events that
take place inside the classroom, it is difficult to be certain that the method-
program distinctions actually result in different classroom behaviours (see
Long 1984). The psychometric approach is also evident in correlational stud¬
ies that have examined the relationship between specific classroom behavi¬
ours (such as teachers’ requests) and learning outcomes (for example, Politzer
1980; Politzer, Ramirez, and Lewis 1981). Chaudron (1988: 30) notes that
these studies suffer from a failure to validate the categories used to measure
instructional features, as well as from a failure to establish theoretical links
between the processes observed and learning outcomes.

Interaction analysis

Interaction analysis involves the use of a form or schedule consisting of a set

of categories for coding specific classroom behaviours. Long (1980b) refers to
three different types of interaction analysis: in a category system each event is
coded each time it occurs, in a sign system each event is recorded only once
within a fixed time span, while in a rating scale an estimate of how frequently
a specific type of event occurred is made after the period of observation. Ini¬
tially, schedules were developed for content classrooms (see Flanders 1970
for an account of this early work), but these were rapidly adapted to the lan¬
guage classroom (for example, Moskowitz 1967). Frequently, the categories
listed in a schedule reflected the researcher’s assumptions about what behavi¬
ours were important and were not theoretically motivated. The Marburg
schedule (Freudenstein 1977) included categories such as ‘phases of instruc¬
tion’, assuming that the basic pattern of a lesson would follow the pattern
‘warming up’, ‘presentation’, ‘learning’, and ‘using’. However, subsequent
schedules (for example, Fanselow 1977a; Allwright 1980) have attempted to
produce more comprehensive and method-neutral sets of categories, while
some (for example, Allen, Frohlich, and Spada 1984) have been based on a
theoretical understanding of the nature of L2 acquisition.2
Long (1980b) lists twenty-two interaction analysis systems to which sev¬
eral more must now be added. This proliferation reflects differences in re¬
search foci. It makes comparison across studies extremely difficult. Also,
because the behaviour of the teacher and the learners is often treated separ¬
ately, information is lost about ‘the sequential flow of classroom activities’
(McLaughlin 1985: 149). Often interaction analysis runs the risk of produc¬
ing disconnected tallies of behaviours that obscure the general picture, as
there is no basis for deciding which combinations of features might be im¬
portant. Interaction analysis depends on a number of assumptions—for ex¬
ample, that it is possible for an observer to ‘read’ the intentions of the teacher
and students. Such assumptions can be questioned, thus casting doubt on the
reliability and validity of the measurements. However, this is more of a prob¬
lem with ‘high inference’ categories (for example, ‘degree of teacher control
568 Classroom second language acquisition

exercised over materials’) than with ‘low inference’ categories (for example,
‘type of materials—text, audio or visual’), as Long (1980b) has pointed out.

Discourse analysis
Discourse analysis serves as a device for systematically describing the kinds of
interactions that occur in language classrooms. Drawing on initial work on
content classrooms by Bellack et al. (1966) and by the Birmingham school of
linguists (Sinclair and Coulthard 1975; Coulthard and Montgomery 1981;
Sinclair and Brazil 1982), discourse analysts give attention not only to the
function of individual utterances but also to how these utterances combine to
form larger discoursal units. They aim to account for the joint contributions
of teacher and student and to describe all the data, avoiding the kind of ‘rag
bag’ category found in many interaction analysis schedules.
McTear (1975) has shown how the Birmingham framework can be ad¬
apted to account for the discourse structure found in language lessons. Re¬
searchers at CRAPEL in the University of Nancy (e.g Gremmo, Holec, and
Riley 1978; Riley 1985) have also made use of discourse analysis to show
how ‘natural’ discourse is distorted in the language classroom as the result of
the teacher’s dominance. Other researchers (for example, Ellis 1980 and
1984a; Van Lier 1982 and 1988) have developed frameworks based on dis¬
course analysis to characterize the different types of interaction that can oc¬
cur in the L2 classroom. More commonly, researchers have used the
techniques of discourse analysis to develop comprehensive accounts of spe¬
cific areas of discourse. Chaudron (1977), for example, provides an analysis
of teacher feedback, while Long and Sato (1983) offer a discourse-based ana¬
lysis of teachers’ questions. In the case of these latter studies, discourse cat¬
egories have served as a basis for quantitative research, but, as Chaudron
(1988) notes, much of the research within the discourse analysis tradition has
been descriptive in nature, concentrating on the development of comprehens¬
ive analytical systems.

r
The ethnographic tradition

The ethnographic tradition involves the kind of detailed descriptive work ad¬
vocated by Van Lier (1988). It emphasizes the importance of obtaining mul¬
tiple perspectives through triangulation. Long (1980b) identifies a number of
ethnographic approaches. Participant ethnography involves the researcher
taking a regular part in the activities under study. Non-participant ethno¬
graphy sets the researcher outside the classroom events being observed. It
uses a variety of data collection techniques: note-taking, interviewing, ques¬
tionnaires, ratings of personal opinions, and written documents (for ex¬
ample, teachers’ handouts and students’ homework). Constitutive
ethnography (see Mehan 1979) aims to make explicit the way in which class-
 Classroom interaction and second language acquisition 569

room participants succeed in creating and managing the events in which they
take part by repeated viewings of videotaped lessons and by soliciting the re¬
actions of the participants to the events that took place. Microethnography,
which focuses on ‘particular cultural scenes in key institutional settings’
(Erickson and Mohatt 1982, quoted in Mitchell 1985) works in a similar way
to constitutive ethnography, but restricts the object of study to contexts of
activity recognized as real by the participants (for example, ‘getting ready for
a break’).
The ethnographic tradition has been particularly evident in research into
bilingual classrooms (for example, Phillips 1972; Trueba, Guthrie, and Au
1981; Wong-Fillmore 1985; Cathcart 1986) and the diary studies discussed
in Chapter 11. Gaies (1983a) gives three advantages of such research: (1) it
can account for learners who do not participate actively in class, (2) it can
provide insights into the conscious thought processes of participants, and (3)
it helps to identify variables which have not previously been acknowledged.
Disadvantages include the time-consuming nature of the work needed to
collect data, the difficulty of generalizing results, and the danger of ignoring
superordinate variables relating to the learners’ social context.
These four traditions have produced a wealth of research into L2 class¬
room processes and products. It is to the results (and insights) which this re¬
search has provided that we now turn.

Comparative method studies

The aim of comparative method studies is to establish which of two or more
methods or general approaches to language teaching is most effective in terms
of the actual learning (the ‘product’) that is achieved after a given period of
time. Many of the earlier studies were ‘global’ in nature, conducted over
weeks, months, and even years. The later ones have tended to examine differ¬
ences resulting from shorter periods of exposure to different methods.
The 1960s were characterized by what Diller (1978) called ‘the language
teaching controversy’. This pitted the claims of rationalist approaches to lan¬
guage teaching against those of empiricists. Methods such as the traditional
grammar-translation method and the cognitive-code method emphasized
the provision of explicit knowledge through rule explanation and of learning
simultaneously through all four skills, while so-called ‘functional’ methods
such as audiolingualism and the oral approach emphasized inductive rule
learning through listening and extensive oral practice. At the time, it seemed
logical to investigate which method produced the better results.
In an early study, Scherer and Wertheimer (1964) compared the grammar-
translation method and the audiolingual approach by following the progress
of different groups of college-level students of L2 German taught by each
method and tested at the end of the their first and second year of study. The
results showed that students in the grammar-translation group did better in
570 Classroom second language acquisition

reading and writing while the students in the audiolingual group did better at
listening and speaking. In other words, each method resulted in learning
‘products’ that reflected the instructional emphasis.
A subsequent large-scale study known as the Pennsylvania Project (Smith
1970) compared the effects of three methods on beginning and intermediate
French and German classes at the high-school level. The three methods were
(1) ‘traditional’ (i.e. grammar-translation), (2) ‘functional skills’ (essentially
the audiolingual approach), and (3) ‘functional skills plus grammar’. Student
achievement in the four skills of listening comprehension, speaking, reading,
and writing was evaluated at mid-year and at the end of the year using a bat¬
tery of standardized tests. The results in general showed no significant differ¬
ences between the three methods, except that the ‘traditional’ group was
superior to the other two groups on two of the reading tests. After two years,
the ‘traditional’ group again surpassed the ‘functional skills’ group in reading
ability but did significantly worse on a test of oral mimicry. No differences
were found in the students’ performance on the other tests.
Hauptman (1970) compared the effects of instruction based on a ‘struc¬
tural’ approach (where the grammatical structures were sequenced according
to level of difficulty) and a ‘situational’ approach (consisting of dialogues ar¬
ranged by situation). In this case the learners were third to sixth grade stu¬
dents of Japanese. In general, the situational group performed better than the
structural group on situational test items, and did as well on structural test
items; however, this difference was found to be significant only for high apti¬
tude/high IQ students in the situational group. Design problems (such as the
short length of the teaching programme—only three weeks—and the fact
that one person did all the teaching) make it difficult to reach firm conclusions
from this study.
In general, therefore, these studies failed to provide convincing evidence
that one method was superior to another. One possible explanation, as sug¬
gested by Clark (1969) in his detailed discussion of the Pennsylvania Project,
was that the distinctions between the different methods were not in fact clear.
Clark comments:

If ostensibly different teaching methods tend in the course of the experi¬

ment to resemble one another in terms of what actually goes on in the class¬
room, the likelihood of finding significant differences in student
performance is accordingly reduced.

As Allwright (1988) has carefully documented, this led to the development of

instruments for observing what actually takes place during the course of in¬
struction (for example, Jarvis 1968) and a focus on describing language les¬
sons at the level of technique rather than method (for example, Politzer
1970).
Comparative method studies were not abandoned, however. Asher and as¬
sociates (for example, Asher, Kusudo, and de la Torre 1974; see Asher 1977
 Classroom interaction and second language acquisition 571

for a review) conducted a number of studies designed to compare the effects

on learning of Total Physical Response (TPR), a method devised and pro¬
moted by Asher, and other methods, in particular the audiolingual approach.
Asher claims that the results of these studies show that TPR results in greater
short-term and long-term retention of new linguistic material and better un¬
derstanding of novel utterances. He also suggests they indicate that learners
are able to transfer the training in listening they received to other
skills—speaking, reading, and writing. Furthermore, Asher reports that stu¬
dents taught by TPR were more likely to continue studying the foreign lan¬
guage and displayed more positive attitudes. Krashen (1982: 156) concludes
his survey of Asher’s studies by claiming that ‘the TPR results are clear and
consistent, and the magnitude of superiority of TPR is quite striking’. A num¬
ber of caveats are in order, however. First, Asher had a vested interest in find¬
ing in favour of TPR; second, the period of instruction was relatively
short—only 20 hours in some studies; third, only beginners were investig¬
ated. We do not know, therefore, whether the method is equally effective in
the long term or with advanced learners. Nevertheless, the TPR studies stand
out in comparative method studies as providing evidence that support the su¬
periority of a particular method.
More recent discussions of language teaching methodology have emphas¬
ized the importance of providing opportunities for learners to communicate.
The few studies that have investigated the effectiveness of communicative
language teaching, however, have led to the same kind of indeterminate res¬
ults found in the early global method studies. Palmer (1979) compared the ef¬
fects of ‘traditional’ instruction and ‘communicative’ instruction involving
extensive peer-communication based on language games. The subjects were
Thai learners of English. No significant differences between the groups were
found. One possible explanation for this unexpected finding, according to
Krashen (1981), was that while teacher talk was in the target language in the
traditional group, it was in the learners’ LI in the communicative class. Ham¬
mond (1988) compared groups of students in a Spanish program at two uni¬
versities. Eight experimental groups were taught by means of the Natural
Approach (Krashen and Terrell 1983) and 52 control groups were taught by
means of the grammar-translation method, which emphasized the deductive
learning of grammar. Although in general the experimental groups outper¬
formed the control groups in both a mid-term and a final examination, many
of the differences were not significant. This study did show, however, that
students in a communicative classroom did no worse in learning grammar
than those in a traditional program—a point we will take up in a later section.
The difficulties of conducting effective comparisons between ‘communic¬
ative’ and ‘non-communicative’ classrooms are also evident in a carefully
planned study by Allen, Swain, Harley, and Cummins (1990). This was based
on Stern’s (1990) distinction between ‘experiential’ and ‘analytic’ teaching
strategies (see Table 13.2). Allen et al. used an interaction analysis schedule
572 Classroom second language acquisition

(called the Communicative Orientation to Language Teaching, or COLT) to

describe the teaching strategies used in eight grade 11 French classes in
Toronto. The information provided by this was used to rank the classes ac¬
cording to how experiential/analytic they were. This study differs from the
earlier ones, then, in that the comparison involved examining classrooms that
varied in terms of the actual processes observed to take place (in other words,
it had a process element). It was hypothesized that the analytic classes would
perform better on the written and grammatical accuracy measures and that
the experiential classes would do better on measures of sociolinguistic and
discourse competence, but when pre- and post-test measures were compared,
this was not found to be the case. There were a few statistically significant dif¬
ferences between the two most experiential and the two most analytic classes,
but even in this comparison Allen et al. admit the results were ‘somewhat dis¬
appointing’.

Experimental features Analytic features

1 Substantive or motivated topic or theme 1 Focus on aspects of L2, including

(topics are not arbitrary or trivial).
2 Students engage in purposeful activity
(tasks or projects), not exercises.
3 Language use has characteristics of real
talk (conversation) or uses any of the four
skills as part of purposeful action.
4 Priority of meaning transfer and fluency
of over linguistic error avoidance and
accuracy.
5 Diversity of social interaction.
phonology, grammar functions,
discourse, sociolinguistics.
2 Cognitive study of language items (rules
and regularities are noted; items are
made salient, and related to other items
and systems.
3 Practice of rehearsal of language items
or skill aspects.
4 Attention to accuracy and error
avoidance.
5 Diversity of social interaction desirable.

Table 13.2: Experiential and analytic features in language pedagogy (from Stern
, 1990)

With the exception of the TPR studies, then, comparative method studies
have failed to produce evidence that one method results in more successful
learning than another. There are many reasons for this. As Lightbown (1990)
notes in her discussion of Allen et al.’s study, one is that foreign language
lessons of any type often result in relatively little progress. Another is that
individual learners benefit from different types of instruction (a point taken
up in the next chapter). A third reason is that language classes tend to offer
very similar opportunities for learning irrespective of their methodological
orientation. It is probably true to say that comparative method studies, even
when conducted with due regard for classroom ‘processes’, afford little
insight into how instructional events contribute to learning. ‘Method’ may
 Classroom interaction and second language acquisition 573

not be the most appropriate unit for investigating the effect that language
teaching has on L2 learning.

Aspects of classroom interaction

The perceived failure of the global method studies of the 1960s led to the de¬
velopment of an alternative approach to investigating classroom language
learning. According to Gaies (1983a), classroom process research was based
on three basic premises; a rejection of the notion that classrooms differ on a
single variable such as ‘method’, an emphasis on describing instructional
events as fully as possible as a way of generating (rather than testing) hypoth¬
eses, and the priority of direct observation of classroom lessons. Following
work on content classrooms, the perceived goal was to discover and describe
how teachers ‘accomplish classroom lessons’ (Mehan 1974). Classroom pro¬
cess research viewed language lessons as ‘socially constructed events’ and
sought to understand how they took place. As Allwright (1983) notes, the re¬
search was ‘illuminative’ rather than hypothesis-testing.
Initially, the research was not informed by any underlying theory of L2 ac¬
quisition, but attention focused increasingly on examining instructional
events in relation to the kinds of interactional theories discussed in Chapter 7.
In particular, Long’s claims about the value of interactionally modified input
has supported a number of studies investigating the extent to which the nego¬
tiation of meaning occurs in different instructional contexts. Some of these
studies continued to rely on the direct observation of actual classroom events,
but others sought to manipulate instructional variables such as participant
(as in studies of small group work) and task (as in studies of different types of
pedagogic tasks) experimentally.
No attempt will be made here to follow the historical progression of class¬
room-orientated L2 research. Instead, we will begin with research that has
considered the general nature of L2 classroom discourse, and then move on to
examine studies that have investigated a number of different aspects of class¬
room interaction (for example, teacher talk and teachers’ questions).

The nature of second language classroom discourse

Classroom discourse mediates between pedagogic decision-making and the
outcomes of language instruction, as Figure 13.1 from Allwright and Bailey
(1991) shows. Teachers plan their lessons by making selections with regard to
what to teach (syllabus), how to teach (method), and perhaps also the nature
of the social relationships they want to encourage (atmosphere). When acted
on, their plans result in ‘classroom interaction’. This is not planned in ad¬
vance, but rather is ‘co-produced’ with the learners. In part, it will reflect the
pedagogic decisions that have been taken, but it will also evolve as part of the
process of accomplishing the lesson. The interaction provides learners with
574 Classroom second language acquisition

opportunities to encounter input or to practise the L2. It also creates in the

learners a ‘state of receptivity’, defined as ‘an active openness, a willingness to
encounter the language and the culture’ (op cit.: 23).

Input

Practice opportunities

Receptivity

Figure 13.1: The relationship between plans and outcomes (from Allwright and
Bailey 1991: 25)

Our concern here is with the -nature of classroom interaction. Although

classroom interactions are not usually designed in advance of a lesson, they
have been found to manifest distinct and fairly predictable characteristics.
These will be considered in terms of (1) the structure and general character¬
istics of classroom discourse, (2) types of language use, (3) turn-taking, and
(4) differences between classroom and naturalistic discourse.

Structure and general characteristics

Classroom discourse has an identifiable structure. Mehan (1979) distin¬

guished three components in general subject lessons: (1) an opening phase,
where the participants ‘inform each other that they are, in fact, going to con¬
duct a lesson as opposed to some other activity’, (2) an instructional phase,
where information is exchanged between teacher and students, and (3) a clos¬
ing phase, where participants are reminded of what went on in the core of the
lesson. However, language lessons often seem just to start and stop (i.e. they
consist entirely of the instructional phase), perhaps because the content is
‘language’ itself.
In one of the best known accounts of classroom discourse, Sinclair and
Coulthard (1975) develop a hierarchical model by identifying the following
‘ranks’ in the structure of a lesson: (1) lesson, (2) transaction, (3) exchange,
(4) move, and (5) act. Overall, a ‘lesson’ has only a weakly defined structure,
consisting of ‘an unordered series of transactions’. A ‘transaction’ consists of
a ‘preliminary’, one or more ‘medial’, and a ‘terminal’ exchanges. It is most
easily identifiable by means of ‘boundary exchanges’, signalled by framing
and focusing moves. The element of structure that is most clearly defined,
however, is that of ‘teaching exchange’, which typically has three phases,
involving an ‘initiating’ move, a ‘responding’ move, and a ‘follow-up’ move,
as in this example:

T: Ask Anan what his name is? (initiating)

S: What’s your name? (responding)
T: Good, (follow-up).
 Classroom interaction and second language acquisition 575

This exchange became known as ‘IRF’. Each move is realized by means of

various kinds of ‘acts’—the smallest unit in the discourse system. For ex¬
ample, the follow-up move can be performed by means of an ‘accept’ (for ex¬
ample, ‘yes’), an ‘evaluate’ (for example, ‘good’) or a ‘comment’ (for
example, ‘that’s interesting’).
This system was not developed to account for language lessons, but as
McTear (1975) has shown, it fits remarkably well. Only small changes are
necessary. For example, students in the F2 classroom often produce an
additional response after the follow-up move in IRF exchanges:

T: What do you do every morning?

S: I clean my teeth.
T: You clean your teeth every morning.
S: I clean my teeth every morning.

The exchange structure is, therefore, IRF(R). It should be noted, however,

that this structure is only likely to arise in classroom discourse which is
teacher-controlled. Although IRF(R) exchanges tend to dominate, other
kinds can also be found. Van Tier (1988) points out that it is easy to overstate
the lack of flexibility evident in F2 classroom discourse. He found that al¬
though the discourse is often strictly controlled by the teacher, learners do
sometimes initiate exchanges, and ‘schismic talk’ (talk that deviates from
some predetermined plan) also occurs, at least in some classrooms.
A number of interaction analysis schedules have also attempted to account
for the general characteristics of classroom discourse. Some of the most fre¬
quently cited are Fanselow (1977a), Allwright (1980) and Allen, Frohlich and
Spada (1984). The last mentioned was used in the comparison of experiential
and analytic teaching approaches discussed in the previous section. We will
focus on it here.
The Communicative Orientation in Fanguage Teaching (COFT) (Allen,
Frohlich and Spada 1984) differs from the systems that preceded it in that it
was not only informed by current theories of communicative competence and
communicative language teaching but also by research into FI and F2 ac¬
quisition.3 The authors comment:

The observational categories are designed (a) to capture significant features

of verbal interaction in F2 classrooms and (b) to provide a means of com¬
paring some aspects of classroom discourse with natural language as it is
used outside the classroom (1984: 232).

The system is in two parts. The first part, ‘A description of classroom activi¬
ties’, is designed for use in real-time coding. It consists of a set of general cate¬
gories broken down into narrower sub-categories, in line with earlier
systems. The main unit of analysis is ‘activity type’; examples of the kinds of
activities that might be identified are drill, translation, discussion, game and
dialogue. Each activity is then described in terms of participant organization
576 Classroom second language acquisition

(whether whole class, group work, or individual work), content (the subject
matter of the activities), student modality (the various skills involved in the
activity), and materials (type, length, and source/purpose). This part, there¬
fore, relies on pedagogic rather than interactional constructs. However, the
second part, ‘Communicative features’, reflects a more discoursal perspective
on the classroom. Coding is based on an audio recording of the classes ob¬
served. Seven communicative features are identified relating to use of the tar¬
get language, whether and to what extent there is an information gap,
sustained speech, whether the focus is on code or message, the way in which
incorporation of preceding utterances takes place, discourse initiation, and
the degree to which linguistic form is restricted. Table 13.3 illustrates Part
Two of the system by comparing extracts from two lessons, one reflecting a
‘stereotyped routine’, and the other communication that is closer to natural
language behaviour. Unlike other interaction analysis systems, COLT has
been used quite extensively in classroom research (Spada 1987; Harley,
Allen, Cummins, and Swain 1990). It has been successfully used to distin¬
guish the types of interaction that occur in different L2 classrooms.

A Analytical instruction

Utterance Communicative features

T: What’s the date today? L2/pseudo-request/minimal speech

Sp April 15th Unpredictable information/ultraminimal speech/limited form
T: Good. L2/comment/minimal speech
T: What’s the date today? L2/pseudo-request/minimal speech
S2: April 15th. unpredictable information/ultraminimal speech/limited form
T: Good. uncomment/minimal speech

B Experiential instruction

Utterance Communicative features

T: t/Vhat did you do on the L2/genuine request/minimal speech

weekend?
S: 1 went to see a movie. L2/giving unpredictable information/minimal speech/
unrestricted form
T: That’s interesting. What L2/comment/elaboration (genuine request for information)/
did you see? sustained speech
S: E.T. 1 really liked it. He’s L2/giving unpredictable information/sustained speech/
so cute. unrestricted form
T: Yes. 1 saw it too and L2/comment/expansion/elaboration (genuine request for
really liked it. Did information)/sustained speech.
anyone else see it?

Table 13.3: Interaction in analytical and experiential type lessons (from Allen,
Frohlich, and Spada, 1984)
 Classroom interaction and second language acquisition 577

The aim of interaction analysis systems such as COLT is to identify signific¬

ant aspects of L2 classroom discourse and to develop specific categories that
allow for quantification. By so doing they provide a basis for investigating
which interactional features are important for language acquisition. How¬
ever, as we noted earlier in this chapter, they have a number of drawbacks.

Types of language use

Other researchers have sought to describe classroom interaction by identi¬

fying the different types of language use or interaction found in L2 class¬
rooms. We will now examine a number of such accounts.
Allwright (1980) provides what he calls ‘a macro-analysis of language
teaching and learning’ by identifying three basic elements:

1 Samples, instances of the target language, in isolation or in use

2 Guidance, instances of communication concerning the nature of the tar¬
get language
3 Management activities, aimed at ensuring the profitable occurrence of
(1) and (2) (1980: 166).

These elements are not mutually exclusive, as instances of ‘guidance’ and

‘management activities’ automatically provide ‘samples’. They are held to
vary according to their relative proportion, their distribution between teacher
and learner, their sequencing, and the language used (target or other).
Other accounts of types of classroom use distinguish between interaction
where the focus is the code itself (a key feature of the language classroom) and
interactions which centre on genuine meaning exchange. McTear (1975), for
instance, identifies four types of language use based on this general
distinction:

1 Mechanical (no exchange of meaning is involved)

2 Meaningful (meaning is contextualized but there is still no information
conveyed)
3 Pseudo-communicative (i.e. new information is conveyed but in a man¬
ner that is unlikely in naturalistic discourse)
4 Real communication (i.e. spontaneous speech resulting from the ex¬
change of opinions, jokes, classroom management, etc.).

Mechanical and meaningful language use involve a focus on the code, while
real communication by definition entails genuine information exchange;
pseudo-communicative lies somewhere in between.
The frameworks developed by myself and Van Lier are a little more com¬
plicated, as they involve two dimensions rather than one. In Ellis 1984a, I dis¬
tinguished ‘goal’ (the overall purpose of an interaction) and ‘address’ (who
talks to whom). Three goals were specified: (1) core goals, where the focus is
on the language itself (medium), on some other content (message), or
578 Classroom second language acquisition

embedded in some ongoing activity such as model-making (activity); (2)

framework goals associated with the organization and management of class¬
room events; and (3) social goals. I discussed interactional sequences taken
from an ESL classroom in Britain to illustrate how the type of goal influences
the discourse, and then speculated about the learning opportunities each type
affords. I pointed out that interactional events with core goals are likely to re¬
strict learners to a responding role, whereas framework and social goals pro¬
vide opportunities for them to initiate discourse and to perform a wider range
of language functions. Kaneko (1991) used this framework to examine the ef¬
fect of language choice (target or the learners’ LI) on learning. She found that
the items that Japanese high school students reported learning in English les¬
sons occurred most commonly in interactional events with core goals where
there was an element of spontaneous language use.
In Van Lier’s (1982; 1988) framework, there are four basic types of class¬
room interaction, according to whether the teacher controls the topic (i.e.
what is talked about) and the activity (i.e. the way the topic is talked about).
Type 1 occurs when the teacher controls neither topic nor activity, as in the
small talk sometimes found at the beginning of a lesson or in private talk be¬
tween students. In Type 2 the teacher controls the topic but not the activity; it
occurs when the teacher makes an announcement, gives instructions, or de¬
livers a lecture. Type 3 involves teacher control of both topic and activity, as
when the teacher elicits responses in a language drill. In Type 4 the teacher
controls the activity but not the topic, as in small-group work where the pro¬
cedural rules are specified but the students are free to choose what to talk
about. In a further development of this framework, Van Tier (1991) adds a
third dimension: the function that the language serves. He follows Halliday
(1973) in distinguishing three types of function: ideational (telling people
facts or experiences), interpersonal (working on relationships with people),
and textual (signalling connections and boundaries, clarifying, summarizing,
etc.).
These broad frameworks have been developed through ethnographic stud¬
ies of language classrooms and, as such, were not intended to serve as schemes
for coding specific classroom behaviours. Often it is not clear what unit of
classroom discourse they are intended to relate to—in Ellis (1984) I con¬
sidered them in terms of interactional ‘sequences’, loosely defined as a unit of
discourse with a unitary topic and purpose, while Van Lier illustrates his
framework with reference to instructional activities. Their advantage is that
they provide a tool for understanding classroom interaction and how it might
affect learning. Their disadvantage is that they do not permit precise quanti¬
fication, and thus cannot be easily used in experimental or correlational
 Classroom interaction and second language acquisition 579

research that seeks to establish relationships between classroom processes

and L2 learning statistically.

Turn-taking

Research which has specifically examined turn-taking in the L2 classroom

has drawn extensively on ethnomethodological studies of naturally occurring
conversations (for example, Sacks, Schegloff, and Jefferson 1974). These
identified a number of rules that underlie speaker selection and change: only
one speaker speaks at a time; a speaker can select the next speaker by nom¬
inating or by performing the first part of an adjacency pair (for example, ask¬
ing a question that requires an answer); a speaker can alternatively allow the
next speaker to self-select; and there is usually competition to take the next
turn. Classroom researchers frequently highlight the differences between
turn-taking in natural and classroom settings. McHoul (1978), for instance,
has shown that classroom discourse is often organized so that there is a strict
allocation of turns in order to cope with potential transition and distribution
problems and that who speaks to whom at what time is firmly controlled. As a
result there is less turn-by-turn negotiation and competition, and individual
student initiatives are discouraged.
Turn-taking in language classrooms does not differ from that in general
subject classrooms. Lorscher (1986) examined turn-taking in English lessons
in different types of German secondary schools and found that turns were al¬
most invariably allocated by the teacher, the right to speak returned to the
teacher when a student turn was completed, and the teacher had the right to
interrupt or stop a student turn. Lorscher argues that these rules are deter¬
mined by the nature of the school as a public institution and by the teaching-
learning process.
The most extensive discussion of turn-taking in the L2 classroom can be
found in Van Lier (1988). In an attempt to identify participation in classroom
discourse characterized by learner initiative, Van Lier identifies a number of
turn-taking behaviours that he considers indicative of such initiative. Ex¬
amples, under the headings provided by Van Lier, are:

Topic

The turn is off-stream (i.e. discontinuing), introduces something new, or

denies/disputes a proposition in a previous turn.
580 Classroom second language acquisition

Self-selection

Selection originates from the speaker.

Allocation

The turn selects one specific next speaker.

Sequence

The turn is independent of sequence.

Van Lier goes on to provide an example of how this system can be used to
assign a ‘participation index’ for each of the participants in an interaction. He
uses this type of analysis to challenge McHoul’s claim that ‘only teachers can
direct speakership in any creative way’ (1978: 188), pointing out that in his
L2 classroom data many of the teacher’s utterances were undirected and the
learners frequently did self-select. He also notes that explicit turn-taking
sanctions are rare in adult L2 classrooms and that there is a considerable de¬
gree of tolerance of unintelligibility. It would seem, therefore, that the rigid
rules of turn-taking described by Lorscher are not always evident.
Studies of turn-taking in the L2 classroom make explicit comparisons be¬
tween classroom and naturalistic discourse, the underlying assumption being
that the former is less conducive to successful L2 acquisition than the latter.
Van Lier offers an interesting explanation for why this might be the case. He
suggests that if turn-taking is rigidly controlled, the learners have no need to
attend carefully to classroom talk to identify potential transition points when
they can take their turn. As a result, they lack ‘an intrinsic motivation for
listening’ and so are robbed of input. However, to date there have been no
studies investigating the effect of different types of turn-taking on acquisition.

The difference between classroom and naturalistic discourse

The discourse that results from trying to learn a language is different from
that which results from trying to communicate. Edmondson (1985) draws on
Labov’s idea of the Observer’s Paradox (see Chapter 4), to suggest that there
is also ‘the teacher’s paradox’, which states:

We seek in the classroom to teach people how to talk when they are not be¬
ing taught (1985: 162).

Thus, there is a tension between discourse that is appropriate to pedagogic

goals and discourse that is appropriate to pedagogic settings. However, be¬
cause the classroom affords opportunities to communicate as well as to learn,
there are ‘co-existing discourse worlds’. Kramsch (1985) suggests that the na¬
ture of classroom discourse will depend on the roles the participants adopt,
the nature of the learning tasks, and the kind of knowledge that is targeted.
Instructional discourse arises when the teacher and the students act out
 Classroom interaction and second language acquisition 581

institutional roles, the tasks are concerned with the transmission and recep¬
tion of information and are controlled by the teacher, and there is a focus on
knowledge as a product and on accuracy. Natural discourse is characterized
by more fluid roles established through interaction, tasks that encourage
equal participation in the negotiation of meaning, and a focus on the inter¬
actional process itself and on fluency. One way in which the two worlds can
be brought together is through communicating about learning itself, as
suggested by Breen (1985).
However, although the potential exists for natural discourse to occur in the
classroom, studies show that it seldom does. Pica and Long (1986) found that
there was very little negotiation of meaning in elementary ESL classrooms in
Philadelphia in comparison to native speaker-non-native speaker conversa¬
tions outside the classroom, as evident in significantly fewer conversational
adjustments by the teachers. Politzer, Ramirez, and Lewis (1981) report that
90 per cent of all student moves were responses, testifying to the limited na¬
ture of opportunities to participate that learners are afforded in class¬
rooms—a point we will take up later. These and other studies (see Glahn and
Holmen 1985; Kasper 1986) testify to the restricted nature of pedagogic dis¬
course, although other studies, such as Enright (1984) show that there can be
considerable variation between classrooms.
The teacher’s control over the discourse is the main reason for the preval¬
ence of pedagogic discourse. Researchers at CRAPEL (for example, Gremmo,
Holec, and Riley 1977; 1978) have argued that in the classroom setting dis¬
course rights are invested in the teacher. It is the teacher who has the right to
participate in all exchanges, to initiate exchanges, to decide on the length of
exchanges, to close exchanges, to include and exclude other participants, etc.
When teachers elect to act as ‘informants’ or ‘knowers’ (Corder 1977b), they
are likely to make full use of their rights, and as a consequence the learners are
placed in a dependent position. As a result there is a preponderance of teacher
acts over student acts (typically in a 2:1 ratio), because teachers open and
close each exchange.
In the opinion of some, pedagogic discourse constitutes a ‘falsification of
behaviour’ and a ‘distortion’ (Riley 1977), but other researchers see it as inev¬
itable and even desirable (for example, Edmondson 1985). To date, there are
more arguments than evidence, although as we will see in Chapter 14, formal
instruction does appear to result in faster learning and higher levels of ulti¬
mate achievement.

Teacher talk
The bulk of L2 classroom research has focused on specific aspects of interac¬
tion, which we will now consider, beginning with teacher talk. Chaudron
(1988: Chapter 3) provides a comprehensive survey of studies of teacher talk.
His main conclusions are summarized in Table 13.4.
582 Classroom second language acquisition

Feature Main conclusions Main studies

Amount of talk In general, the research confirms the finding for Legaretta 1977;
LI classrooms—namely, that the teacher takes Bialystok et al. 1978;
up about two-thirds of the total talking time. Ramirez et al. 1986.
Functional There is considerable evidence of variability Shapiro 1979;
distribution among teachers and programs, but the general Bialystok et al. 1978;
picture is again one of teacher dominance in Ramirez et al. 1986.
that teachers are likely to explain, question and
command and learners to respond.
Rate of speech Teachers, like native speakers in general, slow Henzl 1973;
down their rate of speech when talking to Dahl, 1981; Wesche
learners in comparison to other native speakers and Ready 1985;
and also do so to a greater extent with less Griffiths 1990 and
proficient learners. However, there is 1991a.
considerable variability among teachers.
Pauses Teachers are likely to make use of longer Downes 1981;
pauses when talking to learners than to other Hakansson 1986;
native speakers. Wesche and Ready
1985.
Phonology, There have been few studies which have Henzl 1973 and 1979;
intonation, attempted to quantify these aspects of teacher Downes 1981;
articulation, talk, but teachers appear to speak more loudly Mannon 1986.
stress and to make their speech more distinct when
addressing L2 learners.
Modifications in Several studies provide evidence of a lower Henzl 1979;
vocabulary type-token ratio and teachers also vary in Mizon 1981.
accordance with the learners’ proficiency level,
but Wesche and Ready (1985) found no
significant vocabulary modifications in
university lectures to L2 learners.
Modifications in There is a trend towards shorter utterances Pica and Long 1986;
syntax with less proficient learners, but some studies Gaies 1977;
which use words per utterance as a measure Kleifgen 1985;
report no modifications. The degree of Early 1985; Wesche
subordipation tends to be lower, but again and Ready 1985.
9
results have been mixed. Teachers use fewer
marked structures such as past tense. More
declaratives and statements than questions are
used in comparison to natural discourse.
Ungrammatical teacher talk is rare.
Modifications in There is some evidence that teachers use more Hamayan and Tucker
discourse self-repetitions with L2 learners, in particular 1980;
when they are of low level proficiency. Ellis 1985d.

Table 13.4: Main features of teacher talk (summarized from Chaudron 1988:
Chapter 3)
 Classroom interaction and second language acquisition 583

The research indicates that teachers modify their speech when addressing
L2 learners in the classroom in a number of ways and also that they are sensit¬
ive to their learners’ general proficiency level. Many of these modifications
are the same as those found in foreigner talk, but some seem to reflect the spe¬
cial characteristics of classroom settings—in particular the need to maintain
orderly communication. However, Chaudron does not consider teacher talk
sufficiently different to justify calling it a distinct sociolinguistic register.
Tutor talk (the talk addressed by classroom learners to other, less proficient
learners in the context of peer tutoring) has been studied by Flanigan (1991).
She studied pairs of non-native speaker elementary school children, where the
more linguistically competent child was asked to assist the less competent in
how to use a computer in a graded reading and listening ‘station’. She found
that little negotiation of meaning took place, as the less proficient learners
lacked the ability to respond. However, the more proficient children made use
of the same discourse strategies as those observed in adult caretakers and
teachers: repetitions, expansions, explanations, rephrased questions, and
comprehension checks. Interestingly, the child non-native speaker tutors
made no attempt to simplify their talk grammatically or lexically.
Teacher talk has attracted attention because of its potential effect on learn¬
ers’ comprehension, which has been hypothesized to be important for L2 ac¬
quisition (see Chapter 7). However, little is known about what constitutes
optimal teacher talk. It is not even clear on what basis teachers make their
modifications. Hakansson (1986) speculates that they may aim at some hypo¬
thetical average learner, in which case the input is not likely to be tuned very
accurately to the level of many of the learners in some classrooms. This sug¬
gests that tutor talk may be more facilitative of acquisition than the teacher
talk that occurs in lockstep teaching. Indeed, Flanigan hypothesizes that it
may provide ‘an optimum environment for successful language learning in
the elementary school’ (1991: 153).

Error treatment
There is now a considerable literature dealing with error treatment (see Hen¬
drickson 1978; Chaudron 1987 and 1988; Allwright and Bailey 1991 for sur¬
veys of the literature). Much of this literature is taken up with addressing
whether, when, and how errors should be corrected and who should correct
them. There have been very few studies that have examined the effect of error
treatment on acquisition. In this section we will attempt to clarify the main
issues that have been identified in the research rather than provide a compre¬
hensive survey.
The first issue is terminological in nature. A number of terms have been
used to refer to the general area of error treatment; these are ‘feedback’, ‘re¬
pair’, and ‘correction’. ‘Feedback’ serves as a general cover term for the in¬
formation provided by listeners on the reception and comprehension of
584 Classroom second language acquisition

messages. As Vigil and Oiler (1976) have pointed out, it is useful to distin¬
guish ‘cognitive’ and ‘affective’ feedback; the former relates to actual under¬
standing while the latter concerns the motivational support that interlocutors
provide each other with during an interaction. ‘Repair’ is a somewhat nar¬
rower term used by ethnomethodologists such as Schegloff, Jefferson, and
Sacks (1977) to refer to attempts to identify and remedy communication
problems, including those that derive from linguistic errors. ‘Correction’ has
a narrower meaning still, referring to attempts to deal specifically with lin¬
guistic errors; it constitutes an attempt to supply ‘negative evidence’ (see
Chapter 10) in the form of feedback that draws the learners’ attention to the
errors they have made.
In an attempt to give greater precision to the term ‘treatment’, Chaudron
(1977) distinguishes four types:

1 Treatment that results in learners’ ‘autonomous ability’ to correct them¬

selves on an item.
2 Treatment that results in the elicitation of a correct response from a
learner.
3 Any reaction by the teacher that clearly transforms, disapprovingly re¬
fers to, or demands improvement.
4 Positive or negative reinforcement involving expressions of approval or
disapproval.

Type (1) cannot be determined within the context of a single lesson as it re¬
quires evidence that the feedback has had some effect on acquisition. Type (4)
is limiting because it restricts the object of enquiry to occasions when the
teacher draws explicit attention to learner performance. Most studies have
examined (2) and (3).
A second issue concerns learners’ attitudes towards error treatment. Cath-
cart and Olsen (1976) found that ESL learners like to be corrected by their
teachers and want more correction than they are usually provided with.
Chenoweth et al. (1983) found that learners liked to be corrected not only
during form-focused activities, but also when they were conversing with nat¬
ive speakers. This liking for correction contrasts with the warnings of
Krashen (1982) that correction is both useless for ‘acquisition’ and dangerous
in that it may lead to a negative affective response. Krashen may be partly
right, though, as Cathcart and Olsen also report that when a teacher at¬
tempted to provide the kind of corrections the learners in their study said they
liked, it led to communication which the class found undesirable.
Of considerable interest is the extent to which teachers should correct
learners’ errors. Here, there are widely diverging opinions, reflecting the dif¬
ferent theoretical arguments about the need for negative feedback (see the
section on ‘Error evaluation’ in Chapter 2). Irrespective of what teachers
should do, there is the question of what they actually do do. Chaudron (1988)
reviews a number of studies which have investigated this (Salica 1981;
 Classroom interaction and second language acquisition 585

Courchene 1980; Chaudron 1986; Fanselow 1977b; Lucas 1975). The main
conclusions are that certain types of errors are much more likely to be treated
than others: discourse, content, and lexical errors receive more attention than
phonological or grammatical errors; that many errors are not treated at all;
that the more often a particular type of error is made, the less likely the
teacher is to treat it; and that there is considerable variation among teachers
regarding how frequently error treatment takes place. Edmondson (1985)
has also pointed out that teachers sometimes correct ‘errors’ that have not in
fact been made!
Another issue concerns who performs the treatment. Studies of repair in
naturally-occurring conversations have shown a preference for self-initiated
and self-completed repair (see Chapter 7). In classroom contexts, where, as
we have seen, discourse rights are unevenly invested in the teacher, other-
initiated and other-completed repair can be expected. This is what Van Lier
(1988) predominantly found. Other patterns of repair can also occur, how¬
ever. Kasper (1985) found that in the language-centred phase of an English
lesson in a Grade 10 Danish gymnasium, the trouble sources were identified
by the teacher but they were repaired either by the learners responsible for
them or by other learners. In the content phase of the same lesson, self-
initiated and self-completed repair was evident, although the learners were
inclined to appeal for assistance from the teacher. As Van Lier (1988: 211)
points out, the type of repair work is likely to reflect the nature of the context
which the teacher and learners have jointly created. Also, the teacher’s per¬
sonal teaching style may influence the type of repair work that occurs, as Nys-
trom’s (1983) study of the correction styles of four teachers in bilingual
classrooms demonstrates.
Probably the main finding of studies of error treatment is that it is an
enormously complex process. This is evident in the elaborative decision¬
making systems that have been developed (Long 1977; Day et al. 1984; Chau¬
dron 1977) and also in the extensive taxonomies of the various types of teach¬
ers’ corrective reactions (Allwright 1975; Chaudron 1977; Van Lier 1988).
Chaudron’s system, for example, consists of a total of 31 ‘features’ (correct¬
ive acts that are dependent on context) and ‘types’ (acts capable of standing
independently). These descriptive frameworks provide a basis for examining
teachers’ preferences regarding types of error treatment. Studies have shown,
for instance, that repetitions of various kinds are a common type of corrective
feedback (Salica 1981; Nystrom 1983). The frameworks are indicative of the
careful descriptive work that needs to be done before it is possible to set about
investigating the effects of different kinds of error treatment on acquisition.
Further evidence of the complexity of the decision-making process during
error treatment is the inconsistency and lack of precision that teachers mani¬
fest. Long (1977) notes that teachers often give more than one type of feed¬
back simultaneously, and that many of their feedback moves go unnoticed by
the students. Teachers are likely to use the same overt behaviour for more
586 Classroom second language acquisition

than one purpose. A teacher repetition can occur after a learner error and
serve as a model for imitation, or it can function as a reinforcement of a cor¬
rect response. Teachers often fail to indicate where or how an utterance is de¬
viant. They respond positively even when the learners continue to make the
error. They correct an error in one part of the lesson but ignore it in another.
They may give up on the task of correction if learners do not seem able to
cope. Nystrom (1983) sums it all up with this comment:

teachers typically are unable to sort through the feedback options available
to them and arrive at the most appropriate response.

According to some (for example, Allwright 1975), inconsistency is inevitable

and even desirable, as it reflects the teacher’s attempts to cater for individual
differences among the learners, but according to others (for example, Long
1977), it is damaging. These differences in opinion reflect the relative em¬
phasis given to ‘cognitive’ and ‘affective’ aspects of feedback.
Given the amount of research that has been devoted to error treatment, the
lack of studies that have investigated its effect on acquisition is disappointing.
Chaudron (1986) reports that only 39 per cent of the errors treated in the im¬
mersion classroom he studied resulted in successful student uptake (i.e. were
eliminated in the next student utterance). This might suggest that error treat¬
ment is often not successful, but it is possible that by raising learners’ con¬
sciousness it contributes to acquisition in the long run (see Chapter 14 for a
discussion of the possible delayed effect of formal instruction). Ramirez and
Stromquist (1979) report a positive correlation between the correction of
grammatical errors and gains in linguistic proficiency, but this study did not
examine the relationship between the correction and elimination of specific
errors.
At the moment, therefore, opinions abound about what type of error treat¬
ment is best, but there is little empirical evidence on which to make an in¬
formed choice. Currently, two recommendations seem to find wide approval.
One is that error treatment should be conducted in a manner that is compat¬
ible with general interlanguage development (for example, by only correcting
errors that learners are ready to eliminate). The other is that self-repair is
more conducive to acquisition than other-repair, as it is less likely to result in
a negative affective response (Van Lier 1988). These recommendations await
empirical support.

Teachers’ questions

Teachers, whether in content classrooms or in language classrooms, typically

ask a lot of questions. For example, in three hours of language-content teach¬
ing, Johnston (1990) observed a total of 522 questions of various types. Long
and Sato (1983) observed a total of 938 questions in six elementary level ESL
lessons. One reason for the prevalence of questioning is undoubtedly the
 Classroom interaction and second language acquisition 587

control which it gives the teacher over the discourse. Thus, a question is likely
to occupy the first part of the ubiquitous three-phase IRF exchange. Ques¬
tions typically serve as devices for initiating discourse centred on medium-
orientated goals, although they can also serve a variety of other functions.
Much of the work on questions has centred on developing taxonomies to
describe the different types. In one of the earliest taxonomies, Barnes (1969;
1976) distinguished four types of questions he observed in secondary school
classrooms in Britain: (1) Factual questions (‘what?’), (2) Reasoning ques¬
tions (‘how?’ and ‘why?’), (3) Open questions that do not require any reason¬
ing, and (4) Social questions (questions that influence student behaviour by
means of control or appeal). Barnes made much of the distinction between
two types of reasoning questions: those that are closed in that they are framed
with only one acceptable answer in mind, and those that are open because
they permit a number of different acceptable answers. Barnes also points out
that many questions have the appearance of being open, but, in fact, when the
teacher’s response to a student’s answer is examined, turn out to be closed; he
calls these pseudo-questions.
Kearsley (1976) provides an extensive taxonomy of question types based
on conversational data. Long and Sato (1983) made use of this framework in
their study of ESL teachers’ questions, but found it necessary to make a num¬
ber of changes to include new categories to accommodate questions not ac¬
counted for by Kearsley’s categories, and to eliminate other categories not
exemplified in their classroom data. Their taxonomy can be found in Table
13.5. It centres on the distinction between echoic questions, which ask for the
repetition of an utterance or confirmation that it has been properly un¬
derstood, and epistemic questions which serve the purpose of acquiring in¬
formation. The latter type include referential and display questions, which
Long and Sato discuss in some detail. This distinction is similar but not ident¬
ical to the open/closed distinction of Barnes. Referential questions are genu¬
inely information-seeking, while display questions ‘test’ the learner by
eliciting already known information, as in this example:

T: What’s the capital of Peru ?

S: Lima
T: Good.

Referential questions are likely to be open, while display questions are likely
to be closed, but it is possible to conceive of closed referential questions and of
open display questions.
Other taxonomies have focused on other aspects of teachers’ questions.
Koivukari (1987), for example, is concerned with depth of cognitive pro¬
cessing. Rote questions (those calling for the reproduction of content) are
considered to operate at the surface level, while two kinds of ‘comprehension’
questions (those calling for the reproduction of content and those calling for
the generation of new content) operate at progressively deeper levels.
588 Classroom second language acquisition

Type Sub-category Example

1 Echoic a comprehension All right?; OK?; Does everyone understand

checks ‘polite’?
b clarification requests What do you mean?; 1 don’t understand; What?
c confirmation checks S: Carefully
T: Carefully?;
Did you say ‘he’?
2 Epistemic a referential Why didn’t you do your homework?
b display What’s the opposite of ‘up’ in English?
c expressive It’s interesting the different pronunciations we
have now, but isn’t it?
d rhetorical Why did 1 do that? Because ...

Table 13.5: A taxonomy of the functions of teachers’ questions (from Long and
Sato 1983; based on Kearsley 1976)

Hakansson and Lindberg (1988), in an analysis of questions in Swedish, dis¬

tinguish questions according to form and cognitive level as well as their com¬
municative value and orientation. There are three formal categories: nexus
questions (questions that can be answered ‘yes’ or no’), alternative questions
(that provide the responder with an alternative to select from), and x-ques-
tions (where there is an unknown element, as in w/7-questions). At the cog¬
nitive level, questions are distinguished according to whether they relate to
cognitive memory (i.e. they require some kind of reproduction of informa¬
tion), convergent thinking (i.e. they require the ‘analysis and integration of
given or remembered data within a tightly structured framework’, 1988: 77)
or divergent/evaluative thinking (i.e. they require data to be generated freely
and independently). The categories relating to communicative value reflect
the referential/display distinction, while those relating to communicative
orientation concern whether the question is focused on the language itself
(the medium) or on real-life topics (the message). This latter distinction is
viewed as continuous rather than dichotomous. Hakansson and Lindberg’s
taxonomy is probably the most comprehensive yet devised for the language
classroom.
Whereas there are few problems in assigning teachers’ questions to formal
categories, difficulties do arise with functional, communicative, or cognitive
categories. These are ‘high inference’ and often call for substantial inter¬
pretative work on the part of the analyst. It is not always clear-cut, for ex¬
ample, whether a question is referential or display, as this example from
White (1992) illustrates:

T: How long have you worn glasses ? How long have you had your glasses ?
S: I have worn these glasses for about six years.
T: Very good. Same glasses?
 Classroom interaction and second language acquisition 589

In one sense the teacher’s question is referential as it concerns an area of the

student’s private life she has no knowledge of, but in another sense it can be
considered display, as it is clearly designed to elicit a specific grammatical
structure and was evaluated accordingly. This question suggests the need for
an additional category, such as Barnes’ (1969) ‘pseudo-question’. But it can
only be considered ‘pseudo’ if we know that the teacher intended the question
to be of the display variety—something that we cannot be sure about, as
teachers may ask a referential question only to treat the students’ responses in
the same manner as they would their responses to a display question. The
above exchange is certainly an example of what McTear (1975) called
‘pseudo-communication’. Its prevalence in many classrooms makes the
coding of individual questions problematic to a degree not always recognized
by researchers.
Studies of teachers’ questions in the L2 classroom have focused on the fre¬
quency of the different types of questions, wait time (the length of time the
teacher is prepared to wait for an answer), the nature of the learners’ output
when answering questions, the effect of the learners’ level of proficiency on
questioning, the possibility of training teachers to ask more ‘communicative’
questions, and the variation evident in teachers’ questioning strategies. Much
of the research has been informed by the assumption that L2 learning will be
enhanced if the questions result in active learner participation and meaning
negotiation.
Barnes (1969) found that the secondary school teachers he investigated
showed a clear preference for closed reasoning-type questions. Open reason¬
ing-type questions, where the teacher accepted a number of answers, were
rare. Barnes argues that the teachers saw their role as ‘more a matter of hand¬
ing over ready-made material, whether facts or processes, than a matter of en¬
couraging pupils to participate actively and to bring their own thoughts and
recollections into the conversation’ (1969: 23). Closed questions are indicat¬
ive of a transmission style of education; Barnes argues in favour of a more in¬
terpretative style. Long and Sato (1983) adopt a similar position. They found
that the ESL teachers in their study asked far more display than referential
questions (476 as opposed to 128). This contrasts with native-speaker beha¬
viour outside the classroom where referential questions predominate (999 as
opposed to 2 display questions in the sample they studied). They conclude
that ‘ESL teachers continue to emphasize form over meaning, accuracy over
communication’ (1983: 283-4). Other studies (for example, White and
Lightbown 1984; Early 1985; Ramirez et al. 1986; Johnston 1990; White
1992) also indicate that display/closed questions are more common than re¬
ferential/open questions in the L2 classroom.4
Teachers also seem to prefer instant responses from their students. White
and Lightbown (1984) found that the teachers in their study rarely gave
enough time for students to formulate answers before repeating, rephrasing,
590 Classroom second language acquisition

or redirecting the question at another student. The shorter the wait time,
however, the fewer and the shorter the student responses.
One way in which teachers’ questions might affect L2 acquisition is in
terms of the opportunities they provide for learner output. In Chapter 7 (page
282) we considered the comprehensible output hypothesis, according to
which ‘pushed output’ helps learners to reconstruct their interlanguages. A
key issue, therefore, is whether teachers’ questions cater for such output.
Brock (1986) found that responses to referential questions (mean length = 10
words) were significantly longer than responses to display questions (mean
length = 4.23 words) in four advanced ESL classes at the University of
Hawaii. Similar results have been obtained by Long and Crookes (1987),
Nunan (1990a), and White (1992), suggesting that the findings are fairly
robust. However, as White illustrates, it does not follow that all display
questions produce short responses. It is obvious that ‘there is no meaning left
to negotiate’ in exchanges such as the following:
T: What’s this?
S: It’s a cup.
T: Good.
(Long and Crookes 1987)
However, this is not the case in the following exchange based on a reading
comprehension lesson:
T: Did anyone manage to find some reasons for this?
S: With the decline of religion there is no pressure on woman to get
married.
(White 1992: 26)
In both cases, however, the exchange began with a display question. White
argues that there is a need for a more delicate categorization of display ques¬
tions to allow for the different types of student response evident in these two
exchanges. Banbrook (1987) provides similar examples to show that ref¬
erential questions can also elicit responses of varying lengths and complexity.
It should also be noted that the length of the student’s response to a question
is only one of several possible measures of learner output. Of equal interest,
perhaps, is the length of the sequence initiated by a question. In this respect,
Long and Crookes (1987) report that display questions elicited more student
turns than referential questions. Clearly, more work is needed to tease out the
relationship between question type and learner output.
Very few studies have examined the relationship between teachers’ choice
of questions and the learners’ proficiency level. In Ellis (1985d), I found no
difference in the use a teacher made of open and closed questions with two
learners over a nine month period, but I did find evidence to suggest that the
cognitive complexity of the questions changed, with more questions requir¬
ing some form of comment as opposed to object identification evident at the
end of the period. White (1992) found that one of his teachers used more
 Classroom interaction and second language acquisition 591

referential questions with a high-level class and more display questions with
a low level class, but the other teacher in this study followed the opposite
pattern.
Given the indeterminate nature of the findings of many of the studies, it
might seem premature to prescribe questioning strategies in teacher educa¬
tion. A number of studies, however, have investigated the effect of training
teachers to ask specific types of questions. Brock (1986) and Long and
Crookes (1987) found that instructors given training in the use of referential
questions did respond by increasing the frequency of this type of question in
their teaching. Koivukari (1987) found that training led to teachers using
more ‘deep’ comprehension questions and fewer superficial rote questions,
and was also able to demonstrate that an experimental group who benefited
from this treatment showed improved comprehension scores.
Finally, several studies have pointed to the necessity of acknowledging in¬
dividual variation in teachers’ questioning strategies. White (1992), for in¬
stance, found very different patterns of questioning in his two teachers. Other
studies by Long and Sato (1983), Long and Crookes (1987), Koivukari
(1987), and Johnston (1990) also report extensive differences, although not
all these researchers bothered to draw the reader’s attention to them expli¬
citly. Banbrook and Skehan (1990) provide illustrative evidence to argue that
there is both intra- and inter-teacher variation. They identify three sources of
intra-teacher variation: ‘(a) general teacher variation, (b) variation that takes
place over the phases of the lesson and (c) variation in question asking ... that
is the consequence of the teaching tasks or activities engaged in’ (1990: 150).
They argue that although variation in teachers’ questions is well attested, its
parameters are not yet well understood.
These comments by Banbrook and Skehan lead us back to one of the cent¬
ral issues in L2 classroom research—the extent to which it should be ‘qualitat¬
ive’ and descriptive as opposed to ‘quantitative’ and experimental. Many of
the studies discussed above have belonged to the latter paradigm, but one
might feel along with Van Lier (1988) that:

the practice of questioning in L2 classrooms, pervasive though it is, has so

far received only superficial treatment... An analysis must go beyond sim¬
ple distinctions such as display and referential to carefully examine the pur¬
poses and the effects of questions, not only in terms of linguistic
production, but also in terms of cognitive demands and interactive purpose
(1988:224).

It might also stretch to examining the questions asked by learners. Studies by

Midorikawa (1990) and Robson (1992) suggest that, when learners are given
the opportunity to ask questions, they automatically elect to use open, ref¬
erential-type questions. Rost and Ross (1991) have shown that certain types
of learner question (for example, ‘forward inference’ questions in which the
learner asks a question using information that has already been established)
592 Classroom second language acquisition

facilitate comprehension and, also, that such questioning strategies are train-
able.
Finally, more attention needs to be paid to the socio-cultural context of
questioning strategies. Poole (1992), for example, has suggested that the use
of display or closed questions in most of the classrooms that have been stud¬
ied may reflect the caretaker practices evident in white, middle-class western
society. Teachers from such a background ask questions because they provide
a means by which an expert (the teacher) and a novice (the learner) can jointly
construct a proposition across utterances and speakers. Poole suggests that
this may be why such questions are so ubiquitous and why they are difficult to
get rid of; teachers are being asked to reject a strategy that they feel to be cul¬
turally warranted. A corollary of this line of reasoning is that display ques¬
tions may not be the norm in classrooms where the participants belong to a
culture that does not utilize such caretaking practices.

Learner participation

Given that learners are often restricted to a responding role, it is not surpris¬
ing to find that their opportunities for participating productively in the L2
classroom are constrained. If, as has been hypothesized, opportunities for
using L2 resources are important for acquisition (see the account of the com¬
prehensible output hypothesis in Chapter 7), then it would seem that learning
may be inhibited in the classroom. The assumption that participation is im¬
portant for learning also underlies several of the studies of motivation dis¬
cussed in Chapter 11. We will consider participation here both from the point
of view of both (1) quantity and (2) quality.

Quantity of participation

There is no clear evidence that the extent to which learners participate pro¬
ductively in the classroom affects their rate of development. Table 13.6 sum¬
marizes a number of correlational studies that have examined the
relationship between amount of learner classroom participation and L2
achievement/proficiency. The results are mixed. Whereas studies by Seliger
(1977)/ Naiman et al. (1978), and Strong (1983; 1984) report positive cor¬
relations between various measures of learner participation and proficiency,
Day (1984) (in a careful replication of Seliger’s study) and to a large extent
Ely (1986a) found no such relationships. Allwright (1980) also found that the
learner who participated the most in the lesson he analysed was not among
those who showed the greatest advances. As Chaudron (1988) and I (Ellis
1988b) have pointed out, correlational studies of learner participation are
not easy to interpret, as there is no way of telling whether a ‘participation
causes learning’ or ‘proficiency causes participation’ explanation is correct
when a significant relationship is discovered. As we will see in Chapter 15,
 Classroom interaction and second language acquisition 593

0
0 c \ c
0 o c o
0 o
. c ® =!
0
o £ 0 o £ 0 0 a ^
0 Q. o b <8 S 0 .
£ 0 c 0 0
C 0 n o
® ©
o
o
r
2
^
"O
0 c O o °
+- 0 dO >>
e 0 *o
-Q c
0
CL
p
O
b 03
■O >
0 ■«
0 U 0 0
O 0 0
Q. c
o i- 2 o ■s o 0
c/3 5 _ 9 0 a> c 0 03 o 0 JZ o
0 o 0 0 o C 0 ^ O ■
io “ c >
0 c 0
o E Q. CCS > o CL .
m ® .2 0 0 O ‘O ^3 co 8xi
CO « 0-^0 E <0 C 0 C Q 0 ~ 0 .9- £ .„ C
0 C o ® 0 i= r O £ 0 3
c £ C 0 0
O p -C
8 ££ 0 0
Q.
0
i-
o °
"C
OT 9 C o
■Bl ~ 0
0 i C ®
0 >• 0 o E CD ^ c o t -n w
• C o Q. g>R O © 0 o E

s § 5 g E CO 0 ■2> E
CO 0 _
c
c
0
O
8 !•"
0 "8
0 0 O ° O 0 O 3 2 2
3 3
03 0
3 t Q-'-jE — "D 8 0 -* o | g
0 B c e •- E 0 3 0 0 0 0 0
0
tr o 05
I— 0
6 H 3
o .E to X o
| I >O '3
O O O
3 -C D)
O 0
O 0
z
0 2 x)
nr P

o o
=3 +3 *0 ti T3 . -
03 .52 c O c 0 O 0
3 0 <13 0 O
b w
C
E - c
0.0 0 0 ©
~ _0 o
T3 0 O 0 a. ^5
o E 0
0 -C
• 0 E
E
o o .11
0 0 0 03 o 0 -E" 0 P
0 O 0 O 03 0 0 O
§ C c O o 0 °
E o o ■e 8
03 W '0 0 -1* O 0
S'a
C
o -Q C
^ 0
c 0 0 0
</> 4-T 0 0 | 8 O o 3 ^_: 8?
o 0 x: -C
0
E ® 3 E
o o 03 0
3
0 0
^
CL
0 -P
3 O
c o 2 C 0 _ I *
O <5
0 CD Q.
0 N E 3 O 0 I 0 5 f 9 ®
0 O E "i— t
o 8 0
&—
^ I
O 0 0 O 2 | g o ■ o o
2 o 8 o £ > a Q. O O 03 0 O o o o

3
o
~o03 >
+-•
0
J= C T>
sZ
0 O
0 J£ 0 -P 0
■° 0
0
■O o
0 4=
o 0
c .9 TD 3 3 c 0
0)
E
o "8
3 0 T3 0
O o cr
c O
CO
O 0 0 $ o. 0
c 0
o 0 O) $ § 03
'■3 CO 0 0
0 0 .E 0 03 c
0
a 0 05 c
0 o .52 § i o
c 0
M- 0
o C "O o
Q. O V ««
0 p
0
k_
'€
03 C 0
3 0 0 "O
0 C
i— 0
£
co “
0 0 O 0 E
a X} 3 i— 3 ^ o
0
O
o
~o i “O o >
0 -p T3 0
0
a)
s * l
C
S
TJ
R;o 0
0 8^ u - 0
° 8
gf 0 o
3
0
■-
^
0
C
0
C -P
0
P-C 0 VO
3 .0 co O 0 o 0 C
0 § o
R-f
-O c O
O 0 0 8 «
0
o c 0 :p
0
Q- tl 0
0
0 P- E ro 0-
3 a.
> •p 0 o o rr ® 0
5 < Q-
CO .E 0 o o QC O CC 0 2 W
5

CM _g « £
.0 C
°
C Tp3 45
T3 §- O 3
O
2 § £ Q.S 0
03 <- w
c -r aJ
'c ? O ^ o 0C^ ct
3
0 o 0 • —<00
CO 0
_0 03 03 0 0 o- W
0 o _ cr
0 0 O
o 3 -c TD 03 •
0
p: -Q
a> 3 -S'?
<d "2 o
E 2 -g c .E ~o3 0
■_
0 0
t-E §
"S MO » « 03 P — o
3 CO a
c CD C CM SS 05 0
CO CD LU 3 cc C\J LU r^ -E A= 0

CO ■0-
CO
03 03
T— ■»—
r^- *0 CO
rs. CO
03 0 03 _Q
i— i— 00 CD
O) 03 CO
>» 0 0 03
c
*o .03 E o >
3
*0 0 0 >
55 O) z 55 D LU
594 Classroom second language acquisition

when studies investigating the effects of quantity of practice directed at spe¬

cific grammatical structures are considered, there are grounds for believing
that practice does not make perfect, suggesting that the preferred interpreta¬
tion ought to be that proficiency causes participation. That is, the more profi¬
cient the learners are, the more they get to participate.

Quality of learner participation

While the amount of participation may not be a key factor in L2 acquisition, a
stronger case can be made for the importance of high-quality participation, as
we will see in the following sections that deal with tasks and small-group
work. Swain’s (1985) comprehensible output hypothesis claims that acquisi¬
tion is promoted when there are opportunities for pushed output, not just any
kind of output.
One of the factors that seems to determine the quality of learner participa¬
tion in classroom settings is the degree of control the learners exercise over the
discourse. Cathcart (1986) studied the different kinds of communicative acts
performed by eight Spanish-speaking children in a variety of school settings
(recess, seatwork, free play, ESL instruction, playhouse, interview and story¬
telling). She found that situations where the learner had control of the talk
were characterized by a wide variety of communicative acts and syntactic
structures, whereas the situations where the teacher had control seemed to
produce single-word utterances, short phrases, and formulaic chunks. Other
researchers have also found marked differences in the quality of learners’ par¬
ticipation depending on the kind of activity they are involved in. House
(1986), for instance, compared the performance of advanced German learn¬
ers of L2 English in a role play situation, where they had considerable free¬
dom, and a teacher-led discussion. She found that the learners confined
themselves to an ‘interactional core’ in the discussion, failing to use ‘discourse
lubricants’ such as topic introducers (for example, ‘You know...’) and vari¬
ous kinds of supportive and amplifying moves. In contrast, the role-play con¬
versations sounded much more natural. In cases where participation is
strictly controlled, there may be few opportunities for learners to practise
communicative strategies (Rosing-Schow and Haastrup 1982, cited in Hol-
men 1985). This may be one reason why many foreign language learners reli¬
ant on the classroom fail to develop much strategic competence.
Interactions derived from controlled language practice such as pattern
drills are likely to be characterized by strict teacher control. In Ellis (1988b),
however, I emphasized that practice of this kind is still to be considered a ‘so¬
cial event involving personal investment on the part of the learner’ (1988b:
34), and that some degree of negotiation is still involved. I identified a number
of factors that may affect the nature of the interactions in practice sessions,
for example whether the learners’ responses are volunteered or nominated,
the teacher’s policy regarding the distribution of practice opportunities, and
 Classroom interaction and second language acquisition 595

individual learner differences that affect the degree of anxiety experienced.

Gaies (1983b) also argued that negotiation is evident in teacher-dominated
lessons. He studied the kind of feedback that learners provide in tasks that
required the teacher to describe graphic designs and found evidence of con¬
siderable variation among the learners. It is important, therefore, not to over¬
emphasize the restrictive nature of learner participation in teacher-controlled
interaction.

Tasks and interaction

The study of ‘tasks’ has proved to be one of the most productive seams of L2
classroom research. It has been motivated in part by proposals for ‘task-based
syllabuses’ (see Long 1985b; Prabhu 1987; Long and Crookes 1992). These
attempt to specify the content to be taught in terms of a series of activities to
be performed by the students, either with the teacher or in small group work.
The term ‘task’ remains a somewhat vague one, however, as Crookes’ (1986)
very general definition illustrates:

a piece of work or an activity, usually with a specified objective, under¬

taken as part of an educational course, or at work.

In practice it appears to refer to the idea of some kind of activity designed to

engage the learner in using the language communicatively or reflectively in or¬
der to arrive at an outcome other than that of learning a specified feature of
the L2.6 A task, so defined, can be a real-world activity or a contrived, ped¬
agogic activity (Nunan 1989), as long as the process of completing the task
corresponds to that found in discourse based on the exchange of information.
The main research goal has been to uncover how specific variables affect
the interaction that occurs when learners attempt to perform a task. These
variables can usefully be classified into those that relate to the task itself and
those that relate to the participants performing the task. We will consider the
former in this section, reserving the latter to the following section.
One of the main problems facing research into task variables is determining
which variables to investigate. As Pica, Kanagy, and Falodun (forthcoming)
point out, there are a vast number of social and cognitive parameters that may
lead to variation in learner performance. They argue that there is a need for
some kind of ‘task framework’ that can be used to classify and compare dif¬
ferent tasks. The one they provide is based on two general dimensions: (1)
procedures for communicating the information—who holds the informa¬
tion, who conveys it, who requests it, who gives feedback on comprehension,
the direction of the information flow, whether the information-exchange is
required or optional and the degree of preciseness required; and (2) the task
resolution—whether it is divergent or convergent. However, they also re¬
cognize that there are other dimensions that are potentially important, in¬
cluding the linguistic and cognitive requirements of the task and the
596 Classroom second language acquisition

sociolinguistic context in which the task is carried out. Attempts to classify

task variables in this way are helpful, but we are a long way from developing a
taxonomy that is both complete and psycholinguistically justified.
Equally important in the study of tasks is the choice of measures of inter¬
action (the dependent variable). A number of studies have examined the
quantity of interactional modifications found in the negotiation of meaning
(see Chapter 7) on the grounds that these are important for comprehension,
and thereby also for acquisition. One criticism that might be levelled at such
studies is that they ignore important qualitative differences in the way mean¬
ing is negotiated (Ehrlich, Avery, and Yorio 1989; Newton 1991). A number
of studies have also examined to what extent interlocutors provide feedback
that non-native speakers then act on by incorporating corrected features into
their own production. Other aspects of interaction which have been investig¬
ated include the number and size of turns, repair (and, more narrowly, self¬
corrections), speaking speed, reference (exophoric v. anaphoric), topic, in¬
structional input, hypothesizing and procedural moves. However, most of
these have been investigated in only one or two studies, making it difficult to
reach any firm conclusions.
Table 13.7 summarizes the main studies that have investigated task vari¬
ables. From this it can be seen that several studies have examined the effects
on interaction of tasks that involve a one-way exchange of information as op¬
posed to those that require a two-way exchange. Examples of the former in¬
clude giving instructions and telling a personal story, while an example of the
latter is types of communication games in which the participants each hold
part of the information needed to complete the task. The essential difference
lies in whether the exchange of information is optional or required. A number
of studies (Long 1980a; Doughty and Pica 1986; Newton 1991) show that
two-way tasks result in increased negotiation of meaning. Gass and Varonis
(1985a), however, found no difference in the number of indicators of non¬
understanding (a measure of meaning negotiation) in the two types of tasks.
They suggest that the distinction between one-way and two-way tasks is bet¬
ter seen as continuous rather than dichotomous, which ‘makes comparison a
complex process’ (1985a: 159). However, Long (1989) considers the results
of the research sufficiently robust to claim that ‘two-way tasks produce more
negotiation work and more useful negotiation work than one-way tasks’
(1989: 13).
This is one of three conclusions regarding task-types that Long arrives at in
his own survey of the research. The second is that ‘planned tasks “stretch” in¬
terlanguages further and promote destabilization more than unplanned
tasks’ (1989:14). This is supported by task-based studies of learner variability
(see Chapter 4) which show that learners tend to produce more complex and
more target-like language when they have time to plan their output. It should
be noted, however, that it is not yet clear whether performing in planned tasks
subsequently helps learners to perform better in unplanned tasks.
 Classroom interaction and second language acquisition 597

m « CD
03 c « ® c >> .
£ o © 1
0 0 0
0 C/D 0
c 3 Z =3
z o © o -C © TD
o c CD
0 O O o E C
o 0 Q. 0 Q.
TD 4= x=n 5
0 I
c > 0
0 c
3 Q. TD 3 ■E ® o m ©
-Q 0 c O ^0 c -2
c
o
5 E
0 3
C JZ 0 0 c A
03 0 E 3 01 OT
s CD -^3 © 0 A
— 0 CL JZ —
c _C 0 C C © CD
O 1c 0 c U ° §
| c c o "0 0 0.©
0 0 0 C 0
"• 3 C
c ■g CD 0
X CO o O
o > i 0 o o .8 o
&
0 § CO c
c 0 o c E >> 0 C
"D 0 c ■— 0

Studies investigating the effects of task-variables on L2 interaction

o § O °
0) c o :p c C -C > +- 0
r 0 0 +- 6*g 3
td tz 0 * ■g -g 0 • © 2
e-E c o
> CD C £ 3 s |
0 0 0 0 -X 0
o
c 0 0 _C 0 >
3 a> CD O
03 CD O jg
O c
0 £1, E co 0 TD
C
5 7 p
CD “ 0 « 8 ®
CM c == C A 0 CD 0 3
o o C O £ 2 10
C
c ^ 0 3
CO CD © E ® o c 0 9 p m
o © O >X (C
td 71 0 O ^ 0
03 O 0 A o « 0 0 r-
>> c 0 O CD
0 "V T3 C
td 0 C 0 § e a “ ^ ©
■g cd
‘c .y o x s i
c/d c 0 O O 6 o
> 0 0 ts-c ^ *- 0
z 0 0 ^
0
O CD (£l ■-P O D)
z © c c z E O -C C C TD 0
i 0 0 D 0) 5 0
O CL CD d-p o £ ?
cn 0 o 2 c _g
z _g 0 ?°o TD 0
TD c c E 0 Co® 0 0
> c o TD © 0 © .2 _co o TD 0
TD 0 ^
A s & 0 C 3 I— .
(1)
5 »
O 0 C 0 .© 0 C • ^ 0 C 0 <
0 © 8
'A O 0 i- 0 ^ C — P © 8
CD CD C o 8 o CO A
U E Z ■« o | | 0 0
0 0 3 CD 2 a-
S.E

O 0
tl
8
o ©

±z
£
C
03
0 © >,
0
CD
0
C
© 0 © ^
CL
0 ^
s
■o|
0 +=
O O
2 o
® | ^
0 c 0
O © C
jg 8
TD .©
0 0 O O C/D $ © CL O
© 55 CL Z ^ O O
2 C/D Z Z ~ § I

Q. > Q. Q. CM
0 0 0 TD 0 CD
CD CD 0 CD 0 0 © .
C “-m C
0 ^ ■> ^ > > CM £0 0 C
0 0 O ^ CO 7; c ‘3 0
o o 0 C 0
o r O « 8 2 0 _ 0 ? O CL
0 *“ c 0 CD c © CL 3 0 O
CM o 0 £ O TD 0 © d)
E CM o E 0 E C t O
— -♦-> •F 0 CL ^ 0 0 O
wr £ £ o CM O
o o
C *♦= ^ O >/§-
c 0 .2 ra o ^ 0 2 • ? E
jg Q- CD CD
>> C TD P> >, ? O jg
CD 0 o P to <D .£
.£ o <i
O c
0 A
o SI
ra S «
>> N 0 0
CD
0 v CL TD c 0 A: ci
O ill
11
0 O © 0 CD 0 (n > TD
0 <i CD <5
0
CD
0
> O q)
o © >
5S5££
n > • -4_> S, ro ci) — X. z >-9-o
.2 E 3
0
> c c
O 0
£
O
c
0
0
> c-S
- 9 Q-
x 0 > 3 o
Ic «
CL.O A
*c _0 0 > x (D
o CD 0 i m CO
0 A ^ S' O 0 ^ JZ TD
■0> 0
_0 0 TD
Table 13.7:

> O CO, © m 0 7= co ® s >

^ 0 CD 0 0 3©
£ > ! I 0 CXL +3 0 O ■D ®
<0
CD c 0
cl -x
0
0
c 0 cr o
p ^ C 0
^ O "D
^7 ° C
^ O XE?
0 O 5 ZC o O 5 Z- 0 0 ZL. O
O iS

I TD
C/D C
0
0 0 0
TD TD 0 TD
0 TD 0 —I LU
TD 0 > 4_i
0 >> > 0 C/D
TD TD LU c
> td 0 ^ 0
td CO C/D TD 7:3 C/D 0 o
10 0 © C/D Z TD O 0
C/D TD C/D _c z
> y 0 to /ft 0
z >N Q_ 2 0
Z 5, 0
CO | ■° z I O
O TD 3 0 0
© C/D 73 C/D TD C/D eg I C/D O C TD C
z .© z Z C/D Z C/D 0
A zw C/D 03 0
z z z CO 0
3 z z - m _0
03 i— t— Z z CM ©
<0 “I z

LO TD TD 03 y— 03
0 00 C LO C 03 03 03
0 03 0 03
O 0 CO CO CO 03 T— 03
oo -X TD
g C 0 0 03 00 >> CO -X C
03 0 03 03 g o
0 A C
c
Tong-

>* 0 0 ‘c O o 03 £
1984

■D 03 TD 0 o 3= 3 0
g 1
3 C 0 o 3 O CD 0 0
0 0 3
■4-< o z
<0 _J ul 0 > 6 cc Q a CL CD m
598 Classroom second language acquisition

Long’s third conclusion is that ‘closed tasks produce more negotiation

work and more useful negotiation work than open tasks’ (1989:16). An open
task is one where the participants know there is no predetermined solution;
examples include free conversation, debate, ranking activities, and suggesting
preferred solutions to problems. A closed task requires the participants to
reach a single, correct solution or one of a small, finite set of correct solutions;
examples include identifying the differences between two pictures and identi¬
fying the perpetrator of a crime. Studies by Duff (1986) and Berwick (1990)
lend support to Long’s conclusion. Long claims that closed tasks result in
more topic and language recycling, more feedback, more feedback incorpora¬
tion, more rephrasing, and more precision. However, as F. Jones (1991)
points out, more open-ended tasks may afford learners greater interactive
freedom to practise conversational skills such as topic selection and change.
The work so far has helped to identify a number of task variables that affect
participant performance, but it is probably premature to reach firm conclu¬
sions. In particular, it is still not clear which task variables are the most im¬
portant ones, and even less clear what effect combining different sets of
variables has. Also, little is known about how the performance elicited by dif¬
ferent tasks affects acquisition (see Chapter 7). Thus, despite Long and
Crookes’ (1987) claim that ‘it should be possible to build up a multi¬
dimensional classification, organizing tasks in terms of their potential for
second language learning on the basis of psycholinguistically and psycho¬
logically-motivated dimensions’, it is likely to be a long time before such a sys¬
tem becomes available.

Small-group work and interaction

Another area of L2 research of considerable pedagogic interest is interaction

in small groups. Group work is often considered an essential feature of com¬
municative language teaching (see Brumfit 1984). Long and Porter (1985)
summarize the main pedagogical arguments in favour of it. It increases lan¬
guage practice opportunities, it improves the quality of student talk, it helps
to individualize instruction, it promotes a positive affective climate, and it
motivates learners to learn. In addition to these pedagogic arguments, a psy-
cholinguistic justification has been advanced: group work provides the kind
of input and opportunities for output that promote rapid L2 acquisition.
A number of studies have compared the interaction in teacher-centred les¬
sons with that found in group work. Long, Adams, McLean, and Castanos
(1976) report that students working in small groups produced a greater
quantity of language and also better quality language than students in a
teacher-fronted, lockstep classroom setting. Small-group work provided
more opportunities for language production and greater variety of language
use in initiating discussion, asking for clarification, interrupting, competing
for the floor, and joking. Pica and Doughty (1985a), however, found no
 Classroom interaction and second language acquisition 599

difference in the overall quantity of interactional adjustments in a one-way

task performed in a lockstep setting and in small-group work, but when they
replicated this study using a two-way task (Pica and Doughty 1985b), they
did find significant differences. The overall results of these two studies are
shown in Table 13.8. They indicate that there is an interaction between the
participation pattern and the type of task. Thus, group-work only results in
more negotiation of meaning if the task is of the required-information ex¬
change type.
Task Participation pattern
Teacher-fronted Group
n % n %

Optional information exchange 347 49 145 40

Required information exchange 385 45 400 66

% = % of T-units and fragments combined.

Table 13.8: Comparison of modified conversational interaction generated by

optional vs required information exchange tasks on teacher-fronted and group
participation patterns (from Pica and Doughty 1988: 51)

As reported in Chapter 7, several other studies confirm the value of in¬

terlanguage talk as a source of opportunities for meaning negotiation (for ex¬
ample, Gass and Varonis 1985a and Porter 1986). Rulon and McCreary
(1986) investigated the effect of participation pattern on the negotiation of
content, defined as ‘the process of spoken interaction whereby the content of
a previously encountered passage (aural or written) is clarified to the satisfac¬
tion of both parties’ (1986: 183). They found little difference between small-
group and teacher-led discussions with regard to length of utterance, syn¬
tactic complexity, or interactional features, but they did find that significantly
more negotiation of content occurred in the small-group discussions.
From these studies it seems reasonable to conclude that interaction be¬
tween learners can provide the interactional conditions which have been hy¬
pothesized to facilitate acquisition more readily than can interaction
involving teachers. However, a word of caution is in order. Both Pica and
Doughty, and Porter, show, not surprisingly, that interlanguage talk is less
grammatical than teacher talk. It is possible, therefore, as Plann (1977) has
suggested, that exposure to incorrect peer input may lead to fossihzation.
However, two of Porter’s findings give reason to believe that this may not be
the case. She reports that when learners corrected each other’s errors, they did
so wrongly only 0.3 per cent of the time and also that only 3 per cent of the
errors the learners produced could be attributed to repetition of a fellow-
learner’s error. In general, therefore, learners do not appear to be unduly dis¬
advantaged by exposure to deviant input from other learners. In one respect,
though, interlanguage talk may be inferior. Porter found that non-native
speakers did not provide sociolinguistically appropriate input and suggests
600 Classroom second language acquisition

that learners may not be able to develop sociolinguistic competence from

each other—a point that is taken up in the following section.
These studies were guided by Long’s hypotheses regarding the importance
of interactionally modified input. Bygate (1988), however, like Hatch
(1978a), emphasizes the role of learner output in collaborative discourse con¬
struction. He suggests that group work facilitates acquisition by affording
learners opportunities to build up utterances through the use of satellite units.
These are words, phrases or clauses that constitute either moodless utterances
that lack a finite verb or some kind of syntactically dependent unit, as in the
following example:

SI: at the door

S2: yes in the same door I think
S3: besides the man who is leaving
S2: behind him.

Bygate found numerous examples of such units in the speech produced by

learners working in small groups. He argues that the use of satellite units al¬
lows for flexibility in communication, allows the learner time to prepare
messages, and allows messages to be built up collaboratively thus helping to
extend learners’ capabilities. However, as Bygate provides no comparable
data from teacher-controlled lessons, it is not clear whether satellite units are
a special feature of group work. Fasrch’s (1985) examples of vertical con¬
structions in interactions controlled by the teacher suggest that they might
not be, although it is probable that they are at least more frequent in in¬
terlanguage talk.
Other studies of non-native speaker-non-native speaker interaction have
focused on the participant variables that affect the quality of the talk. One
variable that has been found to be important is sex (see Chapter 6). Another
variable that has been investigated is the proficiency level of the learners. Por¬
ter (1986), in the same study referred to above, investigated the input and out¬
put in non-native speaker dyads of mixed proficiency levels. Not surprisingly,
intermediate learners got more input and better quality input from advanced
than from other intermediate learners. Conversely, advanced learners get
more opportunity to practise when they are communicating with intermedi¬
ate learners. Porter feels that mixed pairings offer something to both sets of
learners. In a very careful study involving a task that required the resolution
of a number of ‘referential conflicts’ (the subjects were given maps that dif¬
fered in a number of ways), Yule and McDonald (1990) examined the effects
of proficiency in mixed-level dyads, where in some interactions the sender of
the information was of low proficiency and the receiver high proficiency
(L>H) and in others the opposite (H>L). They found that the L>H inter¬
actions were at least twice as long as the H>L interactions. Furthermore,
negotiated solutions to the referential problems were much more likely to
take place in the L>H condition than in the H>L (a 67.5 per cent success rate
 Classroom interaction and second language acquisition 601

as opposed to 17.5 per cent). Where mixed ability pairings are involved,
therefore, success (and perhaps also acquisition) is more likely if the lower-
level learner is in charge of the key information that has to be communicated.
To sum up, the research suggests that learners will benefit from interacting
in small group work. They will have more opportunity to speak, to negotiate
meaning and content, and to construct discourse collaboratively. This may
help acquisition. They will also be exposed to more ungrammatical input.
Whether this has an adverse effect is not yet clear. It is possible, however, that
small group work does not aid the development of sociolinguistic compet¬
ence. Less is known about the ideal composition of small groups, but studies
suggest that mixed gender and mixed proficiency pairs may be optimal.

Summary

In this section we have considered a number of aspects of L2 classroom dis¬

course. Given the diversity of classroom settings and the breadth of research
now available, it is difficult to arrive at a general picture. The following state¬
ments constitute an attempt to do so, but should be treated circumspectly:
1 L2 classroom discourse often manifests a well-defined structure, con¬
sisting of IRF(R). Different types of language use are evident, reflecting
in particular whether the focus of an interaction is on learning the L2 or
communicating in it. L2 classroom discourse is likely to differ from nat¬
ural discourse in a number of ways, including the rules and mechanisms
for turn-taking.
2 Teachers adapt the way they talk to classroom learners in similar ways to
foreigner talk. They also modify their language in accordance with the
learners’ proficiency level.
3 Error treatment constitutes an interactional event of considerable com¬
plexity. Teachers do not correct all their students’ errors and they are of¬
ten inconsistent in whether they correct and which errors they correct.
Learners often fail to incorporate the teacher’s corrections in their sub¬
sequent responses.
4 Teachers vary considerably in the number and the type of questions they
ask. Several studies have shown that display/closed questions predomin¬
ate over referential/open questions. The latter type may result in more
meaning negotiation and more complex learner output.
5 Studies investigating the relationship between learner participation and
L2 proficiency have produced inconsistent results. Both a ‘participation
causes proficiency’ and a ‘proficiency causes participation’ explanation
need to be considered, with the latter more likely.
6 The quantity and quality of interaction varies according to task. Al¬
though much work needs to be done to establish how specific task char¬
acteristics affect interaction, there is some preliminary evidence to
suggest that tasks which require information exchange, allow learners to
602 Classroom second language acquisition

plan their output, and have a limited number of possible outcomes result
in interaction which facilitates acquisition.
7 Small-group work has been found to provide more opportunities for
meaning negotiation than lockstep teaching, if the tasks are of the ‘re¬
quired-information exchange’ type. The quality of interaction also ap¬
pears to be enhanced if the learners comprising the pair/group are
heterogenous with regard to sex and proficiency level. Group work may
not be the best way to develop sociolinguistic competence.

It needs to be emphasized that the studies considered in this section have been
descriptive in nature. They provide no direct evidence to show any link be¬
tween particular aspects of classroom discourse and L2 learning. In many
cases, however, they have been informed by theories of L2 acquisition, and it
is on the basis of these that claims regarding the facilitative or debilitative ef¬
fects of certain kinds of interaction have been based.

The relationship between classroom interaction and second

language learning
We shall now consider the few studies that have investigated the relationship
between classroom interaction and L2 learning. We will begin with studies
that have examined whether successful L2 learning is possible in a favourable
classroom environment and then move on to look at studies that have tried to
establish direct links between features of interaction and learning.

Second language learning in the communicative classroom

A number of scholars have proposed that the most effective way of develop¬
ing successful L2 competence in a classroom is to ensure that the learners have
sufficient opportunities to participate in discourse directed at the exchange of
information (see Krashen 1982; Swam 1985; Prabhu 1987). According to
this view, the failure of many classroom learners derives from the lack of com¬
prehensible input and/or comprehensible output. One way of investigating
this dlaim is by studying to what extent a communicative classroom environ¬
ment results in successful L2 learning.
There is now convincing evidence that learners can learn ‘naturally’ in a
communicative classroom setting. We have already considered Hammond’s
(1988) study (see page 571). Terrell, Gomez, and Mariscal (1980) showed
that elementary learners of L2 Spanish can successfully acquire various ques¬
tion forms simply as a result of being exposed to questions in the input. 74 per
cent of 7th grade students’ questions and 82 per cent of the 8th and 9th grade
students’ questions were correctly formed even though no explicit instruction
had been provided. Prabhu (1987) developed a programme known as the
Communicational Teaching Project (CTP), which had as its aim the
development of linguistic competence through a task-based approach to
 Classroom interaction and second language acquisition 603

language teaching. This project, which was conducted in a number of

secondary schools in Bangalore and Madras with beginner learners of L2
English, was evaluated by Beretta and Davies (1985). Although not all the
results showed an advantage for the project schools over the control schools,
which were taught by means of the structural-oral-situational method,
Davies and Beretta interpret the results as ‘being, on the whole, positive’ and
conclude that ‘they provide tentative support for the CTP’. Finally, Light-
bown (1992) reports on an interesting project in New Brunswick, in which
Canadian French children in grades 3-6 were taught English by listening to
tapes and following the written text. Results at the end of the third year of the
project showed that ‘students in this program have succeeded in learning at
least as much English as those whose learning had been guided by the teacher
in a more traditional program’ (1992: 362). They were as good even at speak¬
ing English. The program also resulted in very positive student attitudes.
These studies, therefore, demonstrate that the communicative classroom is
effective in promoting L2 acquisition. However, it should be noted that in all
these studies the learners were at an elementary level.
Other studies suggest that communicative classrooms may not be so suc¬
cessful in promoting high levels of linguistic competence. Krashen (1982) has
claimed that immersion classrooms have succeeded in developing very high
levels of L2 proficiency, but there is growing evidence that they are not quite
so successful as Krashen claims. Researchers have for some time recognized
that immersion learners generally fail to acquire certain grammatical distinc¬
tions (see Chapter 7). Other studies also suggest there may be limitations on
what can be achieved in communicative classrooms. Spada and Lightbown
(1989) found that an intensive ESL course (5 hours a day for 5 months),
which was taught by means of communicative methods emphasizing tasks
leading to natural interaction, produced little evidence of syntactic develop¬
ment. For example, the students were only 50 per cent accurate in their use of
plural -s and only 20 per cent in the case of V + -ing. My study of the requests
produced by two classroom learners (Ellis 1992a) also suggests that the com¬
municative classroom may not be well-suited to the achievement of socio-
linguistic competence. I argue that classroom learners may experience
interpersonal needs to perform speech acts such as requests, and also express¬
ive needs that lead them to do so in a varied ways, but that they do not experi¬
ence any sociolinguistic need to modify the way they perform requests in
accordance with situational factors. This is because ‘the classroom consti¬
tutes an environment where the interactants achieve great familiarity with
each other, removing the need for the careful face-work that results in the use
of indirect request-types and extensive modification’ (1992a: 20).
One interpretation of the research on communicative classrooms is as
follows:
604 Classroom second language acquisition

1 Giving beginner learners opportunities for meaningful communication in

the classroom helps to develop communicative abilities and also results in
linguistic abilities no worse than those developed through more tradi¬
tional, form-focused approaches.
2 Communicative classroom settings may not be sufficient to ensure the de¬
velopment of high levels of linguistic and sociolinguistic competence, al¬
though they may be very successful in developing fluency and effective
discourse skills.
The fact that communicating in a language does not guarantee full target-
language competence accords with the findings of other non-classroom stud¬
ies (for example, Schmidt 1983) and with the arguments of Higgs and Clif¬
ford (1982) regarding the limitations of ‘natural’ learning. The problem,
therefore, may not be the classroom per se, but in learners’ general inability to
utilize all the information available to them in the input.

The effect of interaction on acquisition

In Chapter 7 we considered a number of experimental studies that investig¬
ated whether providing learners with the opportunity to negotiate meaning
led to interlanguage development. All of these studies were classroom studies
(see Table 7.5). They suggested that negotiation and contextual support can
aid the acquisition of vocabulary. However, no study to date has shown that
comprehensible input enables learners to acquire grammatical features, al¬
though Nobuyoshi and Ellis (1993: Chapter 6) were able to provide limited
support for the comprehensible output hypothesis.
In addition to these experimental studies, there have been a number of cor¬
relational studies which have looked for a relationship between various as¬
pects of classroom interaction and either learner proficiency or learner output
during instruction. A number of these have already been considered in this
chapter (for example, those investigating the role of learner participation).
Others were considered in Chapter 7. They have produced somewhat mixed
results. Snow and Hoefnagel-Hohle (1982), for instance, found no significant
relationships, whereas Ramirez and Stromquist (1979) and Mitchell, Parkin¬
son, and Johnstone (1981) did find that some measures of interaction (in par¬
ticular those relating to formal language work—for example, teacher’s use of
metalanguage, overt correction, and vocabulary explanation) were correl¬
ated with performance on language tests. One explanation for the mixed res¬
ults can be found in McDonald, Stone, and Yates’ (1977) study of 14 adult
ESL classes in the United States. This showed that different types of classroom
behaviours were related to different kinds of language proficiency. A struc¬
tured teaching style involving teacher-directed practice activities or question-
and-answer sequences was related to improved scores on tests of formal read¬
ing skills, while a more open teaching style characterized by free responses,
games, and discussion was related to improved oral proficiency.
 Classroom interaction and second language acquisition 605

Correlational studies, however, provide only weak, uncertain evidence of the

effects of interaction on learning.
Another approach has been to compare the L2 learning that takes place in
classrooms that differ in a number of major ways. Wong-Fillmore (1982) dis¬
tinguished two basic types of classroom organization in a longitudinal study
of classroom ESL learners in a kindergarten setting: (1) teacher-directed class¬
rooms, where interaction involved the teacher and the class, and (2) learner-
centred organization, where interactions between teacher and individual stu¬
dents and between the learners themselves took place. The classrooms she
investigated also differed with regard to the proportion of L2 learners and
native English-speaking children in them. Wong-Fillmore found that these
two factors interacted to influence L2 learning. Successful language learning
occurred in classes that had a high proportion of L2 learners and were
teacher-directed, and also in classes that were more mixed in composition
(i.e. had more native English-speaking children) and an open organization.
Conversely, much less learning took place in mixed, teacher-directed classes
and open classes with large numbers of L2 learners. This study suggests that
what is important is access to plentiful, well-formed input tailored to the level
of the learner. In classrooms where the learners spend a lot of time talking to
other learners (often in their LI) or where the instructional language is not
properly adapted, little learning is likely. It should be noted, though, that
Wong-Fillmore’s (1982) study was based on a small number of classrooms,
and also that no attempt was made to quantify either the interactional differ¬
ences found in the classrooms or the learning gains.
It can be argued, however, that qualitative research of the kind carried out
by Wong-Fillmore has contributed most to our understanding of what consti¬
tutes an ‘acquisition-rich’ classroom. In a subsequent paper (Wong-Fillmore
1985), she identifies a number of factors that detailed observations of some
60 classrooms have shown to work for language learning. These are grouped
into two sets: (1) class organization and (2) characteristics of teacher talk.
With regard to (1), she reports that ‘the most successful classes for language
learning were the ones that made the greatest use of teacher-directed activi¬
ties’ and, conversely, that ‘classes that were open in their structure and those
that made heavy use of individual work were among those found to be among
the least successful for language learning’ (1985: 24). This conclusion, which
is similar to that of her earlier study, denies the advantages often claimed for
small-group work (see the earlier section in this chapter). The features of
teacher-directed interactions that Wong-Fillmore considers facilitative of ac¬
quisition include formal lessons with clearly marked boundaries, the use of
‘lesson scripts’ (i.e. regular formats), and well-established procedures for
allocating turns. The important teacher-talk characteristics are a clear sep¬
aration of languages (i.e. avoidance of translation), an emphasis on commun¬
ication and comprehension by ensuring message redundancy, the avoidance
of ungrammatical teacher-talk, the frequent use of patterns and routines,
606 Classroom second language acquisition

repetitiveness, tailoring questions to suit the learners’ level of proficiency, and

general richness of language. It is interesting to note that these include many
of the features of foreigner talk which have been hypothesized to facilitate
learning (see Chapter 7).
Qualitative research of the kind conducted by Wong-Fillmore provides a
broad picture of how interaction in the classroom can affect L2 learning, but
it does not shed any light on how specific linguistic features are learnt. This
has been one of the goals of the Lancaster group led by Allwright. The method
they have developed involves asking classroom learners to complete an ‘up¬
take chart’ by recording the items that they think they learnt during a particu¬
lar lesson. The lesson is recorded and transcribed and the researcher then
identifies where the uptaken items-occurred during the lesson. This provides a
means for exploring the interactional conditions that might have caused
learners to uptake items. The main study to date is Slimani (1989). This indic¬
ated that neither learner participation nor negotiating meaning led to uptake.
What seemed to have an effect was listening to other students. Items ‘top-
icalized’ by other students—rather than by the teacher—were most likely to
figure as ‘uptake’. However, it is not possible to generalize on the basis of this
study, which examined only a few lessons in one setting. Also, little is known
yet about the reliability and validity of ‘uptake’ as a measure of L2 learning.

Summary

In general, the research that has investigated the effects of interaction on L2

acquisition has been sparse and disappointing, and few conclusions are pos¬
sible. There is some evidence to support the following:

1 Opportunities to negotiate meaning may help the acquisition of L2

vocabulary (Tanaka 1991; Yamazaki 1991).
2 Pushing learners to reformulate their utterances to make them more target-
like may lead to greater grammatical accuracy in the long term (Nobuyoshi
and Ellis 1993).
3 Tfeacher-controlled ‘pedagogic discourse’ may contribute to the acquisi¬
tion of formal language skills, while learner-controlled ‘natural discourse’
may help the development of oral language skills (McDonald, Stone, and
Yates 1977).
4 Learners need access to well-formed input that is tailored to their own level
of understanding. This can be achieved in teacher-directed lessons with a
clearly-defined structure and by well-adjusted teacher talk (Wong-Fillmore
1982; 1985).
5 Listening to other students in teacher-led lessons may be more important
for learning than direct learner participation (Slimani 1989).

These ‘insights’ have been derived from studies employing very different re¬
search methods and they have been informed by widely varying theoretical
 Classroom interaction and second language acquisition 607

perspectives. It is not surprising, therefore, that they do not fit neatly together
into a sharply-defined, balanced picture.

Conclusion

It is easy to point to the weaknesses of L2 classroom research. Long (1990b)

gives a number of reasons why the findings are not yet ready to be passed on
to teachers: (1) the studies have generally been small-scale, (2) they have
tended to be short-term with the result that little is known about the long¬
term effects of specific classroom behaviours, (3) the findings have tended to
be partial or fragmented in that they have focused on isolated aspects of class¬
room life, and (4) many of the studies are methodologically flawed (i.e. the
analyses have not been tested for reliability or for statistical significance).
Long also criticizes ‘the absence of theoretical motivation’ in many studies,
but this reflects his own preference for a ‘theory then research’ over a ‘re¬
search then theory’ approach, as shown in Long (1985a), and would not be
accepted by some researchers, such as Van Lier/
While it is necessary to recognize the weaknesses of the research to date, it
is also important to acknowledge its achievements. These include (1) a gen¬
eral acceptance of the need to balance ‘external’ accounts of language ped¬
agogy of the kind found in methodology handbooks with ‘internal’ accounts
of what actually happens inside the classroom; (2) the availability of a sub¬
stantial body of descriptive information about specific interactional behavi¬
ours (for example, error treatment and questions); (3) a developing
understanding of how specific variables (for example, the learners’ profi¬
ciency and the participation pattern) affect interaction; (4) the availability of
a broad set of tools for examining classroom interaction that can be used not
only for formal research purposes but also in teacher education and by teach¬
ers who wish to join in ‘the research enterprise’, as Allwright (1988: 257) puts
it; and (5) some insights into how interaction shapes L2 learning.
In short, the work done so far has made a real contribution to improving
language pedagogy but has contributed little to our understanding of how in¬
teraction affects acquisition. This is because, understandably given the short
history of L2 classroom research, the bulk of the work has been descriptive
rather than explanatory, hypothesis-forming rather than hypothesis-testing.
It is not even clear yet how the relationship between interaction and acquisi¬
tion can be effectively researched in the multi-participant setting of the typical
classroom. This is so much more complex than the one-to-one participant set¬
tings studied in LI and naturalistic L2 acquisition research. The methodol¬
ogies that have been used to date (experimental, correlational, qualitative,
and ‘uptake’ tracing) all have their limitations. Probably the future lies in
careful longitudinal studies that focus on both specific classroom behaviours
and individual learners, and which make use of both observational and
introspective data.
608 Classroom second language acquisition

Notes
1 The term ‘black box’ is often used to describe the learner’s mind, or the in¬
ternal mechanisms responsible for learning. It is used by Long (1980b) to
refer to the classroom. As was the case with the learner’s mind, little was
known for a long time about what actually took place inside the ‘black
box’ of the classroom.
2 What distinguishes these interaction analysis schedules from the earlier
ones is the attempt to provide a comprehensive rather than selective ac¬
count of classroom discourse. Not surprisingly, therefore, they require the
analyst to work with recorded lessons rather than through observation of
classroom events in real time.
3 Mitchell, Parkinson, and Johnstone (1981) describe another very compre¬
hensive observational schedule which they used in a series of studies of for¬
eign language classrooms in Scotland. Four basic dimensions were
identified: (1) topic of discourse, (2) language activity, (3) pupil mode of in¬
volvement, and (4) class organization.
4 Chaudron (1988) cites one study (by Bialystok, Frohlich, and Howard
1978) which showed teachers preferring general information questions,
which Chaudron considers ‘potentially referential’. However, general in¬
formation questions can also be of the display/closed types. Johnston’s
(1990) study of content classrooms shows that this is in fact often the case.
It is a mistake, therefore, to associate referential/open questions with in¬
formation exchange and display/closed questions with a focus on language
form.
5 As Chaudron (1988) points out, there are methodological problems with
Seliger’s study which make the results he obtained very questionable. For
example, it is based on only six learners selected from a group of twelve
learners, and misuses correlational statistics.
6 This does not accord with all definitions of ‘task’. In some definitions (for
example, Breen 1987) ‘task’ is seen as referring to form-oriented as well as
meaning-oriented activities.
7 Long does in fact accept that post hoc interpretation of results has a place
in L2 classroom research. What is disputable is his assertion that ‘the lack
of a theoretical motivation for studies greatly reduces the value of their
findings’ (1985a: 166).

Further Reading

There is now an enormous body of L2 classroom research.

M. Long, ‘Bibliography of research on second language classroom processes
and classroom second language acquisition.’ Technical Report No. 2, Center
for Second Language Classroom Research (University of Hawaii, 1985),
provides a list of several hundred.
 Classroom interaction and second language acquisition 609

A number of full-length books provide summaries of the research. The

most comprehensive, but perhaps not the most readable, is
C. Chaudron, Second Language Classrooms (Cambridge University Press,
1988).
L. Van Lier, The Classroom and the Language Learner (Longman (1988),
provides an ethnographic perspective;
D. Allwright, Observation in the Language Classroom (Longman, 1988)
provides an interesting account of the historical development of observa¬
tional research;
D. Allwright and K. Bailey, Focus on the Language Classroom (Cambridge
University Press, 1991) provides an accessible account of much of the re¬
search in a book designed to encourage teachers to become researchers.
There are also some useful collections of research papers:
H. Seliger and M. Long (eds.), Classroom Oriented Research in Second Lan¬
guage Acquisition (Newbury House, 1983).
R. Day (ed.), Talking to Learn: Conversation in Second Language Acquisi¬
tion, (Newbury House, 1986).
G. Kasper (ed.), Learning, Teaching and Communication in the Foreign Lan¬
guage Classroom, (Aarhus University Press, 1986).
Key papers or chapters relating to the various topics covered in this chapter
are:

1 Interaction analysis

D. Allwright, ‘Turns, topics and tasks: patterns of participation in language

learning and teaching’ in D. Larsen-Freeman (ed.), Discourse Analysis in Se¬
cond Language Research (Newbury House, 1980).
P. Allen, M. Frohlich, and N. Spada, ‘The communicative orientation of lan¬
guage teaching’ in J. Handscombe, R Orem, and B. Taylor, (eds.), On
TESOL ’83: The Question of Control (TESOL, 1984).

2 Types of classroom interaction

R. Ellis, Classroom Second Language Development, Chapter 5 (Pergamon/

Prentice Hall International 1984).

3 Teacher talk

S. Gaies, ‘The nature of linguistic input in formal second language learning:

Linguistic and communicative strategies’ in H. Brown, C. Yorio, and R.
Crymes (eds.), On TESOL ’77 (TESOL, 1977).

4 Error treatment

M. Long, ‘Teacher feedback on learner errors: mapping cognitions’ in H.

Brown, C. Yorio, and R. Crymes (eds.), On TESOL ’77 (TESOL 1977).
610 Classroom second language acquisition

5 Teacher’s questions

M. Long and C. Sato, ‘Classroom foreigner talk discourse: Forms and func¬
tions of teachers’ questions’ in H. Seliger and M. Long (eds.), Classroom Ori¬
ented Research in Second Language Acquisition, Newbury House(1983).

6 Learner participation

R. Day, ‘Student participation in the ESL classroom.’ Language Learning

(1984) 34:69-98.

7 Task-based interaction

S. Gass and E. Varonis, ‘Task variation and non-native speaker/non-native

speaker negotiation of meaning’ in S. Gass and C. Madden (eds.), Input in Se¬
cond Language Acquisition (Newbury House, 1985).

8 Small-group work and interaction

P. Porter, ‘How learners talk to each other: Input and interaction in task-
centred instruction’ in R. Day (ed.), Talking to Learn (Newbury House,
1986).

9 The effect of interaction on L2 acquisition

R. Ellis, ‘Learning to communicate in the classroom: a study of two learners’

requests.’ Studies in Second Language Acquisition (1992) 14: 1-23
L. Wong-Fillmore, ‘When does teacher talk work as input?’ in S. Gass and C.
Madden (eds.), Input in Second Language Acquisition, (Newbury House,
1985).
14 Formal instruction and second
language acquisition

Introduction

In previous surveys of the research that has investigated the effects of formal
instruction (for example, Long 1983c and 1988; Ellis 1985a and 1990a;
Larsen-Freeman and Long 1991), the term ‘formal instruction’ has been
understood to refer to grammar teaching. This reflects both the importance
which has been traditionally attached to grammar teaching in language ped¬
agogy, and also the centrality of grammar in SLA research. The focus on
grammar has had both a practical and a theoretical motivation. It has helped
teachers to understand the factors that determine whether instruction is suc¬
cessful, and it has helped researchers to explore a number of issues of import¬
ance for theory building—in particular, the relationship between the
linguistic environment and the learner’s internal processing mechanisms.
It is useful to take a broader look at the role of formal instruction. To this
end, it is helpful to identify a number of general areas of language pedagogy.
Figure 14.1 distinguishes between formal instruction directed at cognitive
goals, where the focus is on developing linguistic or communicative compet¬
ence, and metacognitive goals, where the focus is on the use of effective learn¬
ing strategies. Cognitive goals can be divided into two types, depending on
whether the instruction is language-centred or learner-centred. In language-
centred instruction, the goal is some aspect of phonology, lexis, grammar, or
discourse, where all learners receive the same instruction. In learner-centred
instruction, the instruction is still directed at some aspect of language, but an
attempt is made to match the type of instruction to the learner, so that differ¬
ent learners are taught in different ways. Formal instruction directed at meta¬
cognitive goals is concerned with attempts to train learners to use effective
learning strategies. Overall, then, this framework allows us to address the fol¬
lowing three questions:

1 To what extent does instruction directed at teaching specific linguistic

items or rules work? Do learners learn what they are taught?
2 Does learner-instruction matching result in improved language learn¬
ing?
3 Does learner training enhance learners’ ability to learn from formal
instruction?
612 Classroom second language acquisition

This chapter will seek answers to questions (1) and (2). Learner training has
already been considered in Chapter 12.
language-centred
instruction
(pronunciation,
lexis, grammar, discourse)
cognitive
goals
learner-centred
_ instruction
Formal
(learner-instruction
instruction
matching)

metacognitive
_ goals
(strategy
training)

Figure 14.1: Types of formal instruction

Language-centred instruction
In the previous chapter we considered whether learners were able to develop a
native-like competence as a result of communicating in the L2. We saw that
the favourable opportunities for communicating provided by immersion
classrooms result in near-target levels of discourse and strategic competence,
but do not appear to lead to high levels of sociolinguistic or grammatical com¬
petence. We considered a number of explanations for this finding, one of
which was that fully successful classroom language learning requires formal
instruction. It is important to ask, therefore, whether formal instruction does
result in better learning. It is also important to ask whether some types of for¬
mal instruction work better than others.

The effects of formal instruction on second language learning

There is now a substantial body of research that has investigated whether for¬
mal instruction results in better L2 learning. This research is of four different
kinds. One group of studies has sought to examine whether learners who re¬
ceive formal instruction achieve higher levels of L2 proficiency than those
who do not. A second group has considered the effects of formal instruction
on the accuracy with which learners use specific linguistic items and rules. A
third group has studied whether formal instruction affects the order or se¬
quence of acquisition. A fourth group has investigated to what extent any ef¬
fects for instruction are durable (i.e. whether the effects are short term or long
term).
These studies have used a variety of research methods. Some researchers
have sought to compare classroom learners with naturalistic learners. In such
 Formal instruction and second language acquisition 613

cases, the effects of instruction are demonstrated by showing that there are
differences in the proficiency or in the sequence of acquisition of the two sets
of learners. Such comparisons are not easy to interpret, given that any differ¬
ences may be the result of general differences in the learning environments
rather than of focusing on specific properties of the target language. Other
researchers have designed experiments to study whether instruction directed
at specific features results in their acquisition. These experiments provide
clearer evidence, but they are subject to the problems that arise with this kind
of research in educational settings, in particular the difficulty of controlling
extraneous variables (for example, factors affecting the learners and the
teacher as individuals).
Researchers have also measured learning outcomes in different ways. In
some cases a formal language test has been used (for example, a multiple
choice test or a grammaticality judgement test). In other studies, spontaneous
language use has been elicited by some kind of communicative task and oblig¬
atory occasions for the use of specific features identified. As we shall see later,
it is possible that formal instruction will show an effect in the case of tests but
not in natural production—as is predicted by some theories of L2 acquisition
such as Krashen’s Input Hypothesis (see page 273).
This diversity in research methodology can be justified by the need to con¬
ceptualize the relationship between formal instruction and acquisition in dif¬
ferent ways. Teachers use formal instruction because they want to develop
learners’ general proficiency, to improve the accuracy with which they use
specific features, and to help them acquire new linguistic features. However,
the diversity also makes it difficult to compare the results obtained by differ¬
ent studies and, therefore, to reach firm conclusions. For this reason, we will
first consider each major area of research separately.

The effects of formal instruction on general language proficiency

All the studies reviewed in this section have considered learners’ general profi¬
ciency, measured in a variety of ways. These studies did not investigate
whether formal instruction affected learners’ ability to perform specific lin¬
guistic features.
In one of the first reviews of the literature on formal instruction, Long
(1983c) considered a total of eleven studies that had investigated whether
learners who receive formal instruction achieve higher levels of proficiency
than those who do not. He concluded that six of the studies (Carroll 1967;
Chihara and Oiler 1978; Bfiere 1978; Krashen, Seliger and Hartnett 1974;
Krashen, Jones, Zelinksi, and Usprich 1978) lent support to formal instruc¬
tion. Three studies (Upshur 1968; Mason 1971; and Fathman 1975) indic¬
ated that instruction did not help, while one study (Martin 1980) showed that
exposure without formal instruction was beneficial. Long considered two
other studies to be ‘ambiguous cases’ because, although they produced
614 Classroom second language acquisition

negative results for instruction, this was because they had been wrongly inter¬
preted. For example, Long points out that variables to do with instruction,
socio-economic background, amount of exposure, and parental attitudes
were conflated in Hale and Budar’s (1970) study, with the result that it was
impossible to determine which factors were responsible for the observed dif¬
ferences in proficiency level.
Long’s general conclusion was that ‘there is considerable evidence to indic¬
ate that SL instruction does make a difference’ (1983c: 374). He claimed that
the studies suggested that instruction was advantageous (1) for children as
well as adults, (2) for both intermediate and advanced learners, (3) irrespect¬
ive of whether acquisition was measured by means of integrative or discrete-
point tests, and (4) in acquisition-rich as well as acquisition-poor environ¬
ments.
A study published since Long’s review lends support to these general con¬
clusions. Weslander and Stephany (1983) examined the effects of instruction
on 577 children with limited English proficiency in grades 2 to 10 in public
schools in Iowa. Students who received more instruction did better on the Bi¬
lingual Syntax Measure (Burt, Dulay, and Hernandez 1973—see Chapter 3).
The effects were strongest at the lowest levels of proficiency in the first year of
schooling and diminished in later years, Weslander and Stephany’s results
contrast with those obtained by Hale and Budar (1970). This study, which
Long classified as ‘ambiguous’, found that learners who spent two to three
periods each day in ESL classes ‘were being more harmed than helped’
(1983c: 297).
In general, then, these studies show an advantage for formal instruction
over exposure. Long, in his original review, claimed that this conclusion was
damaging to Krashen’s position on formal instruction (namely, that instruc¬
tion does not contribute directly to ‘acquisition’ and should be limited to a
few ‘learnable’ rules). Krashen (1985: 28-31) responded by arguing that the
studies did not in fact show an advantage for formal instruction per se, but
only that learning in a classroom was helpful for ‘beginners’, who found it dif¬
ficult to obtain the comprehensible input they needed in normal communica¬
tion outside the classroom. In this respect, it should be noted that Weslander
and Stephany (1983) reported that it was the beginners who benefited most
from formal instruction in their study. To protect himself against Long’s con¬
clusion that the studies also showed formal instruction was advantageous for
advanced learners, Krashen argued that the subjects in some of the studies
had been wrongly classified as ‘intermediate’ and ‘advanced’. In a response to
this, Long (1988) pointed out that his other main conclusions (i.e. (1), (3),
and (4) above) were also problematic for the Input Hypothesis and that the
conclusion that Krashen himself had reached in an earlier paper was, in fact,
the right one—‘formal instruction is a more efficient way of learning English
for adults than trying to learn it in the streets’ (Krashen, Jones, Zelinski, and
Usprich 1978:260).
 Formal instruction and second language acquisition 615

These studies—and Long’s and Krashen’s subsequent debate—are based

on comparisons of the relative effects of formal instruction and exposure.
However, many learners—including many of those in the studies Long re¬
viewed—experienced both together. It is conceivable, therefore, that what
works best is some form of combination of the two. This possibility has been
examined directly in three other studies.
Savignon (1972), in a frequently cited study of communicative language
teaching, compared the grammatical and communicative skills of three
classes of learners of French as a foreign language who had received four
hours of instruction per week. The experimental group, which was given an
extra hour of communicative tasks, outperformed the other two groups on a
number of ‘communicative’ measures, but not on ‘linguistic’ measures. This
study, then, suggests that a combination of formal and informal instruction
aids the development of communicative language skills in foreign language
learners.
Spada (1986) sought to establish whether there was any interaction be¬
tween type of contact and type of instruction. She investigated the effects of
instruction and exposure on 48 intermediate-level adult learners enrolled in
an intensive six-week ESL course at a Canadian university. Spada found that
although type and amount of contact appeared to account for variation in
some aspects of the learners’ proficiency before the effects of instruction were
considered, it did not account for differences in the learners’ improvement
during the course. Overall, instruction was more important than contact in
accounting for differences in the learners’ L2 proficiency. She also found evid¬
ence of an interaction between the type of instruction that different groups of
learners received and commented:

... contact positively accounted for differences in learners’ improvement

on the grammar and writing tests when the instruction was more form-
focused, and negatively accounted for differences on those measures when
the instruction was less form-focused (1986: 97).

In other words, those learners who had access to both formal instruction
and to exposure to English showed the greatest gains in proficiency. As Spada
(1987:133) commented, ‘attention to both form and meaning works best’ for
L2 learners.
This conclusion is also given support by another study (Montgomery and
Eisenstein 1985). This compared the gains in proficiency observed in a group
of working-class Hispanic students who, in addition to regular ESL classes
aimed primarily at improving grammatical accuracy, also enrolled in a spe¬
cial oral communication course, involving field trips to sites where they rou¬
tinely needed to communicate in English, and a group of similar learners who
experienced only the regular ESL classes. The two groups were compared for
accent, grammar, vocabulary, fluency and comprehension. Both groups im¬
proved their rating from pre- to post-test, but the group who had experienced
616 Classroom second language acquisition

the oral communication programme showed greater gains in grammar and

accent and were also more successful in passing the ESL course (86 per cent
pass rate as opposed to 57 per cent). Montgomery and Eisenstein propose
that ‘a combination of form-oriented and meaning-oriented language teach¬
ing was more beneficial than form-oriented teaching alone’ (1985: 329). It
should be noted, however, that the group in the joint programme received
more overall instruction and that it might have been the total amount of in¬
struction, rather than the type that accounted for their advantage.
In general, then, there is support for the claim that formal instruction helps
learners (both foreign and second) to develop greater L2 proficiency, particu¬
larly if it is linked with opportunities for natural exposure. Foreign learners
appear to benefit by developing greater communicative skills, while second
language learners benefit by developing greater linguistic accuracy. As I
pointed out in Ellis 1990a, however, there are several reasons for exercising
caution in interpreting the studies considered in this section.
First, many of the studies, like Montgomery and Eisenstein’s, failed to con¬
trol for overall amount of combined contact and instruction. Studies which
show that learners who receive instruction and exposure do better than learn¬
ers with just exposure are, by themselves, not very convincing, as this may
simply reflect their greater overall contact. To counter this objection, Long
(1983c) points out that whereas some studies showed that differences in
amounts of instruction were related to differences in proficiency in learners
with the same overall exposure, the reverse was not the case—differences in
the amount of exposure had no effect on the proficiency of learners matched
for amount of instruction. Long argues that these contrasting results enable
us to be more confident in finding a positive effect for instruction in studies
which failed to control for overall opportunities for learning. However, if
such studies are removed on the grounds of this design flaw, the number of
studies in Long’s original review that unambiguously show that instruction
helps is reduced to two.
Another problem is that the studies do not take account of individual dif¬
ferences in learners. Fathman (1978) found that the informal learners in her
study varied in English oral proficiency much more than the formal learners,
leading her to suggest that some were skilful at learning informally while
others had difficulty. Similar points have been made about differences in the
ability of formal learners (see the section on Language aptitude in Chapter 11,
page 494). It is also possible that motivation functions as an important inter¬
vening variable. As Krashen, Jones, Zelinski and Usprich (1978) recognized,
students who are more highly motivated to learn are more likely to enrol in
classes. The positive effect found for instruction may simply reflect stronger
motivation rather than the instruction itself.
Perhaps the most serious problem, however, is that many of the studies
made no attempt to ascertain what took place in the name of ‘instruction’.
They simply equated formal instruction with the number of years spent in the
 Formal instruction and second language acquisition 617

classroom. As a result, we do not know for certain, whether the instruction

was form-focused or communication-orientated. We cannot be sure whether
the classroom learners did better than the naturalistic learners because of for¬
mal instruction or because of access to comprehensible input.
More convincing is the evidence provided by the three later studies, which
suggest that learners progress most rapidly when they experience both formal
instruction and communicative exposure. It should be noted that this accords
with the findings of the ‘good language learner’ studies (see Chapter 12),
which have indicated that successful learners pay attention to language form
and also seek out opportunities for communicating in the L2. As we will see in
a later section, it is possible that the long-term effectiveness of formal instruc¬
tion is contingent on the availability of opportunities to communicate in the
L2.
The studies referred to in this section are summarized in Table 14.1.

The effects of formal instruction on production accuracy

In this section we turn our attention to studies that have investigated the ef¬
fects of formal instruction on learners’ ability to produce specific linguistic
features accurately. These studies constitute a much more powerful test of the
general question ‘Does formal instruction work?’
A number of studies have investigated whether grammar instruction de¬
fined as quantity of practice directed at specific grammatical features results
in increased accuracy. In Ellis 1984c, for example, I found that three hours of
instruction in wfi-questions did not help a group of 13 child ESL learners to
produce this structure more accurately in a game that was designed to elicit
relatively spontaneous oral questions. However, some of the learners did im¬
prove substantially. I investigated whether this was related to the amount of
practice which individual learners had received during the lessons, and found
that those learners who improved the most had practised the least—further
evidence that participation in classroom interaction may not be the secret to
success (see Chapter 13, page 594). Long (1988) accounts for the failure of in¬
struction to work in this study by suggesting that it was directed at a structure
too far in advance of the learners’ stage of development. He points out that
wfi-questions involve complex syntactical permutations which typically
make them acquired late (see Chapter 3).
In a later study (Ellis 1992b), I found that differences in the accuracy with
which adult beginners of L2 German performed the difficult ‘verb-end’ word
order rule in communicative speech were not accountable for by differences
in the amount of practice in this structure they had received over a six-month
period. However, I did note that many of the classroom learners appeared to
have achieved greater accuracy in the use of verb-end than the purely natural¬
istic learners studied in the ZISA project (see Meisel 1983). It is possible, then,
618 Classroom second language acquisition

C c
® to 2? 8 o o
rs o
0o Cc O O 3 E O

Instruction does not help.

Instruction does not help.
Q.
Q.
X
Q.
X
O 0
Q.O- a
x o. a 11
x 9- 0 0 X <D 0 X 0
0 sz 0 _c I D
■o C
d)
13 •- o •- o
"O5 -C "O C
c
03 0) X)
§ B
m ° o ^
n £
5*2
3 0 0
c
o
0) Q. c 8 Q.
O
□
0i-i. =5u; \-L. U>
0
0
o
Q.
X
0
JZ
0
si o 0
0
_c o 0 0 ® 0 E "OO
0 c c D C c D _C "5* 1 ■E 1 c
O +J o -*J o E -A jj
0 o
■-P 0 13 V-» o 0 Z5 O 0
Results

ZJ o o 3 O 0 3
.E -Q o C o c _c _Q o c C n 0 O > 0 o > o
D 0 0 0 o
E 0 E 0 jz CL JZ CL O 0 0 O
Q. 0 0 0 CL 0 £ a 0 £
w o <2 o o 0 0 o o 0 X o O x o O
0 c
£. -o 9 -o CD jz ■O CD .c LU U JZ uu -o JZ

1 Discrete point test

0 0 0
0 0 0 0 0 0

2 Integrative test
0 0 0 0 0 0

Integrative test
C 0 c 0 C 0
0 0 0
0 o c C 0 o o 0 C
CL 0 CL 0 CL 0
> o 'o > > O
0 CL CL 0 0 0 0 Q.
> 0 > >
0 0 0 0 0 0
0 0 0
0 O) 0 O) O) 0
v_ 0
o 0 0 o O
0 0 O) 0 0 0 0 O)
Data

O) o o O
0 c 0 0 0 c 3 0 0
b b b
jZ T- C\J b b ll T- C\J i- C\J c b

All proficiency levels

0 0 0 0 0 0
0 0 0 0 0 0
> > > ■o > > >

Intermediate and
0 "O
0 0 0 c 0 0 c
>> >* >> 0 >> >. 0
O O O 0 O o O 0
Proficiency

c c c c c c
0 0 0 0 0 0 0 0
advanced

advanced
~ -O
o 0 o o ‘o '~o
JZ 0 o 0
o O o c o o o E
^ 0
Q. Q. Q. O) 0 > Q. Q. Q. 0
0 "P "O
CD C
5 5 — 0 5 5 <

0 0 C
Children—mixed first

CD O) 0 "O 0
0 0 0
Adults—mixed first

D 13 T3 0 0 0
C X
U) O) ■Q
■0O ^0
c c b E
0 0 T30 ■Q
0
"O
0
| 0
0
0
x U
0
X X X 0 CD 1 X
■? »
0 C 0 0
languages

languages

languages

0
0 10 0 E E 0 Z3 1 0 E
Subjects

0 0
0
0 E£ o)
0 0
o) O)
0
CD O O)
0 c
c CD
0 0
c 0 0 0 0 _0 0
-O O =5 Z3 =3 T3 O
0
t± O) O)
TD^ 05 a 3 CD 3 CD O CD 3
C 0 sz c c
Q 0 0
T3 C
0
■Q c 13 0 3 C ■Q
< LU < < 0 < o 0 <

0
0 •o 0 0 0
ESL in United States

0
ESL in United States
Type of classroom

35 _ 0 C O 0 0 0 0
0 T3 "O 0 o o
o V 0
0
0 0 0 0
O) 0 0
0 .-t; o 35 0
0 ^ 55 0 35 35 35
c n ■O
0 ^
■Q 35 T3 ■O ■o
g>=> ■§ 0 0 0 0 0 0
CL 0 .E .•tJ ~o
0 C 0 c ^ 0 C 0 C "c c
0
c cn Z3 3 -C CD 3 C 3 3 3
0)0 'ftO _g 0 0 _c 3 c c c
0 C_ a _i
c
0
O
O) _l _i _i _l _l
0 X 0) Q- c CO 05 05 05 05
0 0 LU co J5 LU LU LU LU LU

CO
h-
O) O
0 CO
O) G)
O)
05 0 CD
Fathman 1975

O T3
h- -a 0 Is- ■o =3
N- CO
Mason 1971

CO c O CD
CD
O)
c
0
cn 2 h- 0 CD O)
'r“ C tz
O)
c c ■o C
E 0 0 0 0 c 0
Study

o co JZ c 0 JZ 0 E
.c 0 +- 0 CD _0 -C
0 0 0
CO JZ E 0 E O) 0 CL
0
O O x m 5 £ I LL 3
 Formal instruction and second language acquisition 619

c TJ 0
o a)
o 0 03
~o
C\J
0 *8
c0
0 0 0 n ^0 c0 g o *g o
*= _0 'c=3 03
>0 c >v
0
=3
o u
O . E C E li„ 0
0 £ Q.
'■p
0 0
_0 ■Q o E
o
O00 zoO +-0
c oo
g §■ ? O
0
T3
r-
o ii "O0 On
c CL c 0 ^ O O £ c/> A T-H
O I Cl CL ft 0 03 03 o

c ■§S O o 0 0
0 •- E s
T3 -9- c o
1 O >1 0 0 0
0o .E0 0D C O _Q
0 0 —
«J ^ O
C
0
£ « ~ 0 _>, ^0 cf s
1.2 03 O o o 03 o o
=3 ^ S2 E 0 g 0 ^ 0

3 1| 0 Q)
i ° 0 t P D P > o=3 iL
0
=3
0
(/) 11 E £
0 O 00 T3C o r~ F
Et 0 0
C O
0
oc
0 O
? P 0 J= _J 0 E
_ 0
° Q_ 03 £ c
^
0
0
Q-
v
CD

O
O
0 ^ 0 S
00 w
0
Q_
"O >v
0^-0 vj
0
0 0 § c-
«8 0
O > E 3 o o §> O
=3 0) o
0 i E'-.l 0 '-P
$O c •a
c 2 £> §
C o
0 r o> a sz
0 C 0 to
B a =J >< 3 8E 0a s
03 eg 0
Q. — 00 ^0
cu o E
—C ^
r1 0
0 CO 0 Q 0 O O E E
o o
G
a o
CO viz m co ^ _ 0
rs,
Jz O O
O 0 > o o

S
0
>
0
_0 ^3
_0
to
R
"O
>»
0 c 0 0
O
o cC 0
0 0 0 0 R
c
0
03 jD 0
'u0 "o0 •2
'g n0i ~G0 £o E E VJ

o 0o 11 0 0
S
Ik.
* o O S' C
a. CL O -o

0
C\J
0
>
0 CO
-oc "g2 LU

0^ g
‘c
i i 0Q. tr
o
0 sz
c 0
U
-C
o o 0 c o c
o<D -n w O
^ ® o i?'s ~o0 Ci3
A X O
3 00
CO < 0 CM

0 0 ^3
0 =) 0 S
C
CO CO 0
~G ~o0 •o
0
0 0
c u
‘c 0 'c 0
m D C Z> O
c 0 _c _c
_i
E _i _1
CO 0 CO CO
CO
til

h-
oo
03
T—I

o
~oCLO
"O co -Q 0 CO
C CO -0
0 03
0
Er? CO e2
o
^r ® * cc
E •-
CO
03
CO
03
TD C
£0 _C0 o5
03 0 0
■o
*o3 0 Q-
0 0 EO ^0 0
CO¬
4-»
CO
CO 5 CO
620 Classroom second language acquisition

that the instruction did have some effect but that this effect was not readily
explicable in terms of amount of practice.
Other studies, however, have produced results more supportive of formal
instruction. Lightbown, Spada, and Wallace (1980) studied the effects of
half-hour grammar lessons on the accuracy with which 175 French speaking
school learners of English judged sentences to be grammatical. In this case the
structures involved were morphological (plural -s, possessive -s, 3rd person
singular -s, copula -s, auxiliary -s, and locative prepositions). With the
exception of locative prepositions, all the structures had been previously
taught. Overall scores improved by 11 per cent from pre- to post-test, a
significant increase given the brevity of the instruction. A control group,
which did not receive any instruction in these features, improved by only 3
per cent. Lightbown et al. (1980) emphasize, however, that the learners may
not have acquired the functions of the various forms, in other words that
acquisition was only partial. Pica (1983; 1985) also produces evidence to
suggest that some grammatical features are performed more accurately if
learners have access to formal instruction. She compared the accuracy with
which three groups of learners (a natural group, a mixed group, and an
instructed group) performed a number of grammatical morphemes in
unplanned speech. The instructed learners performed plural -s more accur¬
ately than the naturalistic learners. However, they performed progressive -ing
less accurately, while no difference between the groups was found for another
feature (articles). Pica explains these interesting results by suggesting that
instruction only aids the acquisition of features which are formally easy to
acquire and which manifest transparent form-function relationships. This
accords with the suggestion made above that the success of instruction may
depend on the complexity of the target structure. Pica makes the additional
point that instruction may be effective in inhibiting (although not preventing
altogether) the use of ungrammatical but communicatively effective construc¬
tions found in pidgins.
Yet other studies have indicated that formal instruction has an effect on
accyracy in planned but not in unplanned production. In both of my studies
referred to above (Ellis 1984c and 1992b), acquisition was measured with
reference to relatively unplanned language use (i.e. when the task did not en¬
courage learners to attend consciously to the target structures). Schumann
(1978a) also found that his learner (Alberto) did not improve the accuracy
with which he spontaneously produced English-negative structures after a
period of instruction, but that he did show a significant improvement in an
imitation test. Kadia (1988) studied the effects of 40 minutes of one-to-one
instruction on a Chinese learner’s acquisition of ditransitive (for example, ‘I
showed him the book’) and phrasal verb constructions such as ‘I called him
up’. She concluded:
... formal instruction seemed to have very little effect on spontaneous pro¬
duction, but it was beneficial for controlled performance (1988: 513).
 Formal instruction and second language acquisition 621

Again, though, it should be noted that the structures that Schumann and
Kadia investigated are complex and typically acquired late. In the Pica study,
instructional effects were evident in unplanned language use when the target
structure was easy to acquire.
Another intriguing possibility is that formal instruction may have a delayed
effect. That is, its effects may become evident only some time after the instruc¬
tion. Empirical evidence for such a hypothesis is limited and difficult to col¬
lect, but the hypothesis is compatible with a number of research findings. As
we will shortly see, attempts to teach specific grammatical structures which
learners are not yet developmentally ready to process may not prove success¬
ful. However, studies such as mine in Ellis 1989b and 1992b indicate that
learners who receive such instruction may still do better than untutored learn¬
ers, suggesting that they may have been able to make use of the information
they received through instruction at a later time. A delayed effect hypothesis,
then, is compatible with the general finding that instruction accelerates learn¬
ing and results in higher proficiency levels (see previous section) even though
learners may fail to immediately learn what they have been taught.
There is also some evidence to suggest that formal instruction can have a
deleterious effect. Felix (1981), in a study of 34 children studying L2 English
in their first year of a German high school, found that drilling in negatives
involving the auxiliary ‘doesn’t’ led to errors in declarative sentences such as
the following:

Doesn’t she eat apples.

Here, ‘doesn’t’ has replaced the more normal ‘no’ as a pre-negator. Eubank
(1987b) also found evidence of unique negative structures in the production
of tutored adult L2 learners of German. In this case, the learners placed the
negator at the end of the sentence, a phenomenon not attested in naturalistic
L2 acquisition. He suggests that this arose when the learners resorted to an
operating strategy (i.e. preserve the basic word order) in order to cope with
the instructional demand that they compose in complete sentences. In other
words, when asked to perform beyond their level, learners simplify and un¬
usual errors occur. Lightbown (1983) showed that francophone learners
over-learnt progressive -ing as a result of teaching. They overgeneralized it,
using it in contexts that required the simple form of the verb, which they had
used correctly prior to the instruction. Pica (1983) also found evidence that
instruction can trigger the oversuppliance of a number of regular morpholo¬
gical features such as past tense -ed and progressive -ing. Weinert (1987) and
VanPatten (1990) also provide evidence to suggest that instruction can im¬
pede acquisition of negatives in L2 German and clitic pronouns in L2 Spanish
respectively. In all these studies the reason given for the failure of the formal
instruction is that it distorts the input made available to the learner and thus
prevents the normal processes of acquisition from operating smoothly.
622 Classroom second language acquisition

Formal instruction can also impede acquisition in another way—by mak¬

ing learners too conservative. Felix and Weigl (1991) asked German high
school students of L2 English to judge sentences such as the following:

1 That John will win the race everyone expects.

2 For him to pass the test would be a surprise.
3 Who did he see pictures of?

All of these sentences are ‘marked’ in some way: (1) and (2) do not follow
canonical English word order and (3) displays prepositional stranding. The
results showed that all the learners, irrespective of the number of years of
instruction, tended to judge these grammatical sentences as ungrammatical.
Felix and Weigl suggest that these marked structures are typically not taught
and that the learners followed the strategy ‘if it hasn’t been taught, you better
believe it’s wrong’. In this case, then, formal instruction made learners ‘overly
cautious’. It should be noted, however, that Felix and Weigl do not report any
comparative data for untutored learners.1 This study is also flawed by the fail¬
ure to include a control group of native speakers. It is not at all clear that
native speakers would judge all the sentences above as grammatical.
The general picture which emerges from the studies we have examined so
far is that formal instruction often does not work, particularly when acquisi¬
tion is measured in relation to spontaneous speech. ITowever, a number of
other studies carried out in Canada suggest that grammar teaching can have
positive effects on learning. These studies have examined the role of formal
instruction in the context of communicative language teaching (i.e. when
opportunities for communicating in the L2 are supplemented with grammar
lessons). A feature of all the studies has been the provision of instruction that
has been carefully planned, in accordance with current pedagogic views
about what constitutes ‘good’ grammar teaching. A second feature is that the
instruction has been extensive.
Harley (1989) devised a set of functional-grammar materials to teach
French immersion students the distinction between passe compose and
impqrfait, which as we have seen is one of the features that is typically not ac¬
quired by immersion learners. She found that eight weeks of instruction res¬
ulted in significant improvement in the accuracy with which the two verb
tenses were used in a written composition, in a rational cloze test and in an
oral interview. The instructional effects were therefore evident in both
planned and unplanned language use. However, a control group sub¬
sequently caught up with the experimental group. Harley suggests that this
might have been because the control group also subsequently received formal
instruction directed at this feature. The Harley study is encouraging for sup¬
porters of formal instruction.
Two other Canadian studies also support this view. White (1991) investig¬
ated the effects of instruction on adverb placement, examining whether it was
successful in eliminating this error made by French learners of L2 English:
 Formal instruction and second language acquisition 623

"'John kissed often Mary (SVAO).

and in teaching these learners to use adverbs between subject and verb, a
position not allowed in French:

"'Jean souvent embrasse Marie (SAVO).

White argues that the latter structure is learnable through positive input (i.e.
through input obtained from communicative exposure), but that avoidance
of the adverb placement error requires negative evidence such as that pro¬
vided by formal instruction. The learners were children in Grades 5 and 6 in
an intensive ESL program. Two weeks of instruction in the use of frequency
and manner adverbs was provided three months into the program. The in¬
structed learners showed significantly greater gains in accuracy in a number
of manipulative tasks (a grammaticality judgement task, a preference task
and a sentence construction task) in comparison to control groups. These
results held true for both use of SAVO and for avoidance of SVAO word
order. White was also interested in whether the learners recognized that
sentences of the kind:

John ran quickly to the end of the street (i.e. SVAPP)

were permitted in English even though they had not been taught this. How¬
ever, the subjects did not learn that SVAPP was possible where SVAO was
not, and White suggests that they made ‘a conscious overgeneralization’ to
the effect that ‘adverbs must not appear between the verb and something
else’. This study indicates that the learners responded to instruction, but that
they were unable to make fine distinctions (i.e. between SVAO and SVAPP)
that they had not been taught.
The second study (White et al. 1991) studied the effects of instruction on
question-formation {tub- and ‘yes/no’) on the same groups of learners as those
used in the adverb study. Five hours of instruction over a two-week period
were provided. Acquisition was measured by means of a cartoon task, a pref¬
erence grammaticality judgement task, and an oral communication task. In
comparison to a control group, the experimental group showed substantial
gains in accuracy in all three tasks. The instructed learners in this case showed
that they had learnt how to use inversion in questions.
The studies considered in this section are summarized in Table 14.2. What
general conclusions can we come to on the basis of the research which has in¬
vestigated the effects of formal instruction on accuracy? There is sufficient
evidence to show that formal instruction can result in definite gains in accur¬
acy. If the structure is ‘simple’ in the sense that it does not involve complex
processing operations and is clearly related to a specific function, and if the
formal instruction is extensive and well-planned, it is likely to work. How¬
ever, if the instruction is directed at a difficult grammatical structure which is
substantially beyond the learners’ current interlanguage, it is likely that it will
only lead to improved accuracy in planned language use, when learners can
624 Classroom second language acquisition

o
X c
x 0
x
3
® "S o
OrE 0 0 O ^ 3 — .C
05
C O 0 0 CO > O ^ 0
rj 0 o _ I ro® 0 .2 0 O CO c 2
8 © s £«|(J
0
CO 0 X X 8-g 0 I q> 0 -9 g
10 0
c
r-Q.CC o 0 - O ? O 0 >-
C 0 P 9 2 £0 Z. 2 § 0 0 > ^ 0
C X X
0
— Q i- 0 C 05 00 E OT-P O 3 05 2 3 0
8»i C
1Z
tp
0
0 0
r- 0
O O
~ .
05 ® c o E a 01_g 05 >4—
0 O p 0
| ot X C w > o 05 Q c o <0 E '0 0 m
0 0 0 0 X 33 3 2 CO 2 CD 2 o O 05 3 0
2 o X
O) 5 CL
® iS -o | i 0^° LOE C C 0 O 0a zdcr is** 8 to
LL u.£ c c 0 O > X
c C 0 <00 *- o3 0 .9 «1 ^ c o is, 0 r-- _ca ■ p H- 5-E X b M_ ~ 0
c ° > < 05 2 © a 0 -p C a 9 0 g
0 E 5
■Z o
3 0Q) X
C
> 0 ■a «S
zap g ® CC | §- xj E 0 ~ZZ
C - w 0 — o X o •- 0 E 0
I c O >> Z' X !z 2 ’0
05 "O E S 0
’P O 0 := 0) 0 0
0 M_ .2 c E
o ° ■- .! 0 3 o.E M © J a 3 it •W
O 0 b o 0)-“ T3
3
E u 5 i E 3 0 0^
CD ^ O
E-o -ST O T 'x -q ” i 0 c ^Jj
>, ra « ? 2 o to S Eo ®| 8 E b t ■§ I
0
9 § 0 ■— TZ, E o > -^ 0 O -D .P
0 to £ -o J « o x O 0 3 0 W O £= t g
DC _E 0 -b 0 .=1 2 CO .9 £ < 5 .b
c Cra o o -O -Q i5 c c 0 0 0 E 0 .£ g ©

X c
c 0 E E
0 E o o o o
X 0 o 2 ' x: 0
o0 CD
X
0 0 o 0 0
0 0 0 0 o
0 _0 0 0 >
X Q.
0 O O
0
C jJ
O 0
X
0
0 ^ o o 0 JxC O XI
3 0 0
0 3 0 x 2 Fh"® «
O o 0 .S’ © CT>
0C -*-*C0 0
CD
c
CD
c
O 0
0
C 05
0 0
Q. 0
W >r
© O O
v.
0 O E b x X x 5 3
0 0 ■D ‘F ©
co 0 E . &o 80 O 8 9 £ ® O C
2 to O 0 §| b £ E E 'x
CO
Q
X'l
CO CD ®
® a
oc 0
p a
QC 0
a. a
co 0
0 c
CD § ^
< ‘5

0 0
O 0
_0
a b0
05 0
L. 0
o E 0 0 0 E
c x
n0
c ^ 0 C C 0 0
> C c
0• 0
-P J0 05 05
C M—
o
0
>
■T "E > 0 0 JO _0
o 1 0 X X i
0
c x e X 0 0 >
0 0 5 0 0 0 o 0 o
n a
o ~ o X X X > X >
'0 0 0 E E c 0 c
b 0 > X '0 X
i- o 0 o o '0
a. ll x 2 O O

CO
0 CO?
0
X 0
i—
0
X 0 © «
0 05 E ®
c _c x X5
0 (0 £
o o
0
0 I x
® co ®
0
0 C _i
I I ®| |
0
k-
cp cp >
o ^ _C
X 0
c 0
O • C LJU^ ® LO
0 0 0 O 0 0 0
= lu ro
0 -q _0 i— 0 0 p CO o
x 0 0 = ox > >% X 3 3
3 ^ Q. X C o
<f) O CO O ^ X 0
0 0 II CO 0 C

0 0
0 ^
0 £= o
o
XJ
0
X
0 .9 c
0
E 00 O o O
o x E
0 c -S3
o c •- o c CO
i- 9 0 0 LU
_0
0
0 c X o O
O
0
0 c
JO D 0 « 0 o
c '0
o 0
X 0 8 £> s £r XI
o
X
0 0
.2
O c 10 c 2 CO CD
o _c 0 "O
0 I?
E o E o
C XJ
-- 0
c
a _1 _J ±3
CO $
CD 8
CO LL c CO
£ LU 0 n ®0
CD LU LU Z5

0
C I o
c 5 CO
0 §o X i CO
05 c i® 05 CO
E _q X) 00 T- 0 X 05
X 3 CO b? X
3 -C f^. 0 S 0
P 05 g>_. 0 0 05 .9>§ .9 §
CO C/5 1- _J 0 LL LL t- _l 1- a. 9- LU
 Formal instruction and second language acquisition 625

00
0 ■o 0
E
i— o *
C
0 0 -O C 0
b o

learners reluctant to generalize

0 0 0
h- 1*1 . 1.E ^ 0 0 70
CO o 0 -o ^ "S ffi N 0 .2 0
H— 0 E T3 0 0 ^ 0 O

beyond what they had been

c £ "O
o 0 c O 3 E 0
o 0 o o E 0 0 C C
0 3 1 0
3 ~o 0 C 2
co .3 c "O .? 0 +-* c c C 0
0 I- £ c C -C o o 0 o 0 O) >
£ >, 0 3 •*- 0 ~ 42 o C ‘0 0
C Q) _
E 0 y
.11 |e 0 3= jt ®
I.
o 42 ° c £
0 •= C Is 0 0 n 0
0 c -C "O 0 0 0'
3 0 0 3 03 » 3/C
o Q.
0 0 ■§
0 i_ 0^0 X) ^3
!'!' C — 0 O
o -o g o E E 0 - - 5 <13 ■§

explicitly caught.
0 c P
Pi £
o V> o § *0 ~ > 0 0 o 0 ro
0 "O 0 >,-J 0
0 0 ~ 03 C JZ
0 03 o
0 C
C/3 £ o-S O 0 oS.o 0 ^00
0 0 c "0 .o' 0
to '0
C O 8-3
Sob 0 O ~ -o .2
1 - 0 -0 3 3 03 1|
0 B 0 03 o 0 c O 0 0 ^ £ -F Q.
i'-S 0 0 -u
o 8 o ■qz c 0 0 p
c 0 0 0 0 O 0 O 0 E E P p ^3
w
1 8
0 T 0 T=
^2
E g 0
3
t— E -2 0 5 c t O <B 3 .E 2 £ C 3 E 0 E
(/> — 3 0 0 0
> b TJ 03 0 -= 0 O 0.2 0
a> E o
•—
« 8
0 O 0 o
0 a 0
? 8. ® 'a c 8 I 0 0= o
^ O.E.E
cc co co _C £3 —1 o J£ O a. .2, o O Q. £ II
-0
c 0
0 0C/3
0
0
■° o F E 0
.Q
+-> E
_c
o 0 00
0
5
> 0 S
0 0 03 03
■o
~£
VVJ
a §1
0
0 ■o N
O
0
i_ c
o 0 3 O o
c Q- -Q OO
0 >
0
o 0 c ■o 0 1—
0 0 CL
0
Q- ^ w co 0 0 0 0 0
0 0 0 0 C Q
0 O
o § O 0 O 0
0
E-P 8 o o . 0~ _ -Q it: 0 C 0
c V-
0 0 E 3 0 0 0
O E
2 c 0 0
S-g E 03 0 0 "0
0 3 E E 0 _C CL 0 0 E .
0 E o 0 E O O 0
Evu 2 0
03 0 o l| 0 O
O a
0-3
co 0
2 $
o) 8
0 0 0 c
O -o O j2
D o o CO .2 0 0 0

0
Q
TO 0 0
CL _0 0
0
0 03 a 0 "o
C i—
0 0 0
3 0 o 5 ■o
3 0 c E
0 0
■§
0
*o 0 8
0 ~o c .si 0 0 c o
o .Q 0 0 . - 0 0
■o e
0 0
c 3 E
0 0 TJ 'U ■O II
0 _0 0 jD 0 0 0 o 0 0 0
o c ^ §- E E E
3 -D
Q. C c £
g.i 03
0 0
0 > 0 0 0 03 03x3
o
i_ I i o S’ 0 O 0 c c 0 0 C
CO 0
CL O o o .a CO Li. *0 £ n

c
0
o 0
o
c E
o i_o
■a 0 c
c E
•O H—
■O
0 o 0 0 0 0
c
E o 0 C 0 S5
o o
0
03
i_
o O 0 CL o S 0
0 0
■“ ^ 0
0 -P* ■o
0 .2
o
0 0 c 3 0 s
O C -D ^ '(/) "O 2 2 0
3 0 0 03 0 0 rC
r^
0
0 0 £
o ■O > 0 03 E i?
n ~o
3 o 3 C Is-
3 0 -O 0 c CO -i O 0 Is-
CO CO 0 < CO 3 O

0 0
c •O 0
o 0
■O c
0
c 0 O C
o JO o C 0
c 0 o
-J o
CO j£ 0
LL ■o 0 o 0 0
0 LL. 0 0 0 -p 0 0
0 0 0 c 0 0
0 -*-> 0 >% -C CO
0 >>
c c
c CO .0 Q_ -Q 0 0
It
0 -2 0 •Q -- 0 o C r- 0 E E
0 E © c TO E s
a £ o ’c
_J >
0 CL_I c 0 0
> 0 o 0 0 CO c O O
O CO a CO 3 LU 3 CO LL 3

h- .Q
03
00 *0
03 Is-
00 03
00 CO CO C
03 03 co
0 T~ 03 0
y— 03
■e ? O c T— >% 0 0
> 0
c 0 -O 0 0 0 Q- O X 03
^
TJ
3
0£ -Q TJ -JZ
0
c
0
03
03
:=
0 03
'0 = o 0

<0 5 LU CC I > T_
626 Classroom second language acquisition

00 0c 20

up
0
TD0 c05 005 i 0
t4—
0£ tr o C0 ic

0 80 o0 0> Ip c 0 J5
■o to .E co
TDC ■O 0£ _c E
0 ■t
l o
o 0c
E
0 0 0 O *o to a Eg cto oc
0 co I 010— "O0
to ~
0C tD E E
O o o 0 « tD
00 v_ 0 Q. 0
0 0 c Q. 0 _® _ to
o 0 ‘c
CU Q)

0 0 o0 ■Oo TD0 E
*o C o ■O ™ <?a
© s 0 3
O E > TJ o 2 O >
3 0 0
3 0 0
</>
CD 0C ■QO) -CCD cc0 O) 00 C O
oc 3 ic _0 ll R o
C TD
0 C
E « E 0
0 _vf 0 _* .
O) CO O) 0
"O3 +-0 "O 0 0
j*: 3-^0
■r* cd
o 00 —• 0
>> O C
~ c ~ C O
0 0 O) 0 <D 'y
-9 0 c ■ 9| §
0 *0 ■c 0 *0 "O
E Q_ o F q.
E * EP
E - °-
to
p
0
0
Q J s 5
'"Bo

<:
o o
Ns *Js
0 0
C
0 0•
c
0
7-30
.E .E
O) Vj O) -

?! ^ C
O
o
0 0 0 0
O +=>
Q. Q_ .E CL .£

c
0 0
"O ■o
3
0 ■*->
0
3

CD CD
■Q *o
c c ■
0 0
CO to
0 0
o ■a ■a
<D 0 0
k-
A 05 05
3 C\J CD
cn CD to

0 0
■o "O
0 0
C c
0 0
a o
c
—I
05 05
LU

0 0
0 T_ 0
■o
*-•
3 ic
>5 05 ic 5
c0
 Formal instruction and second language acquisition 627

pay conscious attention to the structure. It will either have no effect on accur¬
acy in unplanned language use or may result in idiosyncratic deviations as
learners resort to alternative processing operations within the compass of
their current abilities. It is possible, though, that in such cases learners may
store some explicit representation of the feature, which may aid subsequent
acquisition of the feature for use in unplanned discourse.
It is clear from these conclusions that one important factor determining
whether formal instruction results in improved accuracy is the learner’s stage
of development. Instruction may lead to more accurate use of grammatical
structures in communication providing a learner is able to process them. In
other words, there are constraints on learners’ ability to acquire grammatical
structures and, if formal instruction is to be successful, it has to work in
accordance with the internal processes that govern why some structures are
acquired before others.

Formal instruction and the sequence of acquisition

In Chapter 3, we considered the evidence that supports a standard order of
acquisition (for example, a ‘natural’ order based on the accuracy with which
L2 learners perform a set of English morphemes) and a regular sequence of
acquisition (for example, the acquisition of English negatives and interrogat-
ives manifests stages of development involving ‘transitional constructions’ en
route to the target-language constructions). It is of both theoretical interest to
SLA researchers, and of practical importance for language pedagogy, to ask
whether formal instruction can ‘subvert’ the natural order and also whether it
can enable the learner to acquire target language constructions immediately
and so avoid transitional constructions.
Do classroom learners (who are presumed to have received grammatical in¬
struction) manifest a different accuracy/acquisition order from naturalistic
learners? Table 14.3 summarizes the studies that have addressed this ques¬
tion. Studies by Fathman (1978) of EFL and ESL learners, by Makino (1980)
of Japanese learners of English, and by Pica (1983) of Spanish learners failed
to find any significant differences. Turner (1979) found that the accuracy/
acquisition orders manifested by three 18-year-old Spanish learners of Eng¬
lish enrolled in an intensive English language program correlated highly with
each other but not with the order of instruction. Perkins and Larsen-Freeman
(1975) found that the order obtained for twelve Venezuelan students at the
beginning of a course of instruction did not differ significantly from that ob¬
tained at the end. These studies were used to claim the inviolability of the
morpheme order. Sajavaara (1981b), in a study of adolescent EFL learners in
Finland, and also Lightbown (1983), in the study referred to earlier, however,
found differences between the instructed and natural morpheme orders and
those produced by the learners who had received formal instruction.
628 Classroom second language acquisition

■g 0
c - c
0 CM
0 Q. CO 0 0
3 C 03
O T3 O LU >* 0 c C 0
0 C
■Q c E |
.2 r 0 .2 ^ E -o
c 0 0 E o CD 03
0
0 0 C
c
«o >, E
0A
c C 0 _l 3P 'S £ N
0
3 ^ E (2
P ^ m
LL C C
O _I '■= O* 0 0 ■§ S
^ iz CD LL 2 03 O
a o UJ g 0 LU 0 0 0 O “ $
Cfl 0 C I 8
00 CD 3 C g 0 T3 0T3O
0 0
8 2B £ ~ B 03 0 r~ i—
£z C E ofl
z 3 CD o
<D CD cd -n a) i ,s> 0 0 0 0 _ x 0
-1 0 0 20
■o c= -C 0 -C 0 0
T3 #>
'>
2 *2 ■p I P I
o fr 3i M- > P- ^ r
CD O £ o 2 a
© ® 0 O : O 0
<s 'Z
O
0
0 -6
^0 "O0
^ c -C -6 £ C Q)
o E
g o O
CD
B co E ^
in « "O
0 —
w -C _2 o 0 CO
3
C/>
a <8 <■> 2 “ fg b 0
0 b
11
fl
0 CO co o 03 O H— « 2
1 -O
fc C CD 111 0 O
DC o o E 0

CD JD
c X. C
o 0 0 0
2= c
2= iS o A
■O o p O
C a) C 0 ~o
03 CD O 0 0
^ Q-'-g Q. 0
C/3 C/3 Q- 0 . 0
S « 5
03 -*CO C3
0
0

c o 0
o
o js 0 ■§ 2 ^
3 10 i
_T3 CL
U 0 Q--S CD
■B § -o o
*0 CO 0 0 S -C
£ ra © C 0 TO V ^ P
CL I E 0 O CD m
CD g c 3 IB
4->

Q
CD 2 8. .2
I— CO Q_
"0
6
8.1
03 03 $ '5.
0

0
0
~o
0
> 0 E
v_
o 5 0
c 0
T3 A c 2D
o 0 0 0
C
0
o T3
E g T3
0 0
o
i_
o X 03 X
0
Q. c
CD

0 T3 ■O CO
c
c B 0
0 c 03 58 !
A -C 0 0
03 0
03 0 P
0 —'
lc *o O
*2 A E CD
C c 0 O _g
_0 C O ra-g 0
>> g co
C LL _J CD
N | 0 0
0 > CO
O O £ o) O . up
C t 0 0
0 _0 0 o ^ LU ffl
^0 CD 0
A O
0 ^ B
3 CM 2? ^ "O ■ -n © ■
cn T- .2, CO < 03 t- < 0 CD ffl

0 03
LU _g
E CO 0 _i
o ■O > 03
o
!_ 0 X LU
0 c .
V) c 0 0 _J 0 T3 O
03 ® c C o
0 3 E 2 0
LU 0 _0 X
O 0 0 B CL
0 C 0
H- o £ 03
o ~3 Ll
C T3 #g c c
(1) -- 0
a _J _i _l _1 <
>. C/3 LL C s-1 LL LL LL 03
LU 3 £ D LU LU LU 3

"O
3
C
0
0
0
U G? CO 00
o —1 r^
"D 03 03 O 03
0 03 CO
C i- 6- 03
E 0 c 03 0 co
0 0 0 0 0 co
>» -C O 03
.£ CD
■D
3 O
Q.
-* E0 E
A
0
C
c
0 0
4-» 0 P 0 0 o
<0 CL LL
3
1-
’0
03 CL
 Formal instruction and second language acquisition 629

c re
c/) u® >> 2 o ■O
1— ■o *o 03
0-g £
C
0 re o ~o o 0 0
c E o B -o c 0
■o ~ o
E ?c c ^ re re £
re
^
0
o -C
C
rf) Q. o © re o ZJ — 0 0
22
re > 3 O
"E o o 2
3
cr o
— o © O >> t _Q 0 3
5 5
h— re E © 0 re
© 22 — -O
O ir 2= re o re re 0 0
> -Q g- 22 0 ■P' o c o .E 3
2 re c- "O — *o g = c °
re o c > - 2= 0 P re 0 .C
o =j re tj 0) O 03 E
re S E -2 ■O 03 0 o c -E £ 3 £ re
c
3 C .03 0 * « -2 0 3 03 0
C 22
o
— ao ^ g £ 2 .h= re
c Jr 0 0 Lr
-P
c re -o
o
~
3
cr 73
©
f£ ■d 3 E © 0 0 ® o o
=3 0 03 o 0 cr c o> 2 Q. 0
O E TD P
to 0 >* ‘O
re re 8 S e! © m o 0 6 m' C
re £ re -c O re —
c E oE ~
c 0 ^ -*-< i-
o — 0 0
c o
° .E
c- c 3 0 > co <J ^ 0 re
> 0 P re a-E ©
§ S -i=© re 3 O
*- -C -»-» c o T3 c 0 cr
o
0
-2 -c-g "O 0 m C .© 0 E
c
=5
cr C 0) O E re 2-2 & re o *° E -o
Si- -Q re - 0 2= CL o E 0 0 Q. c
§! re re o 0 .2 *o
0^0
0 0 > ^ 0 .O 0
3 3 © .c 0 3
c 0 re — 0 111 0 0 ® p _ § 22
(0 a. £ 0 E "g o a b 4= J= > 0
0 0 C/5
VJ > re re _c o»» ©0 2= i?-e*E o re P_ -D . re 0 ©00
cc £ -o o) c
— 0
£ P3 > O T2 .E 03 T2 TO [-0-0 re c O o 03 .E © -o >

re
»—
3
O
Q.
>, 8|
C © 0 ©
22 re 03
0 ~
s "s t«
§ D)
C ' 0 o ©
o p ■> ° § §
re 0 £
"cQ ©
§ 2 0
re ©
-g 0 .E § C
re - « « c E
•i-1 re 3 "re
re TO 0
—T 2? R|
a O o o o z 6 CO .E

0
0 _0
0
>
C
c
re
•4—
0 03 TJ
0 C
22 ©
o 0 0 0
c 22 22 -♦—>
0 0
0 re re © c c
o "D ■O Q. c c
0 0 E 03 Q3
o X X
o 0 0
Q. O CQ 22

0
-C
0 03
2Z .X
03 21
I C
f! 1^-
.E c ■O
re
0 re
03
c re re
0 ±1
c 0 C
0
1” ^ o
O 5
o -C P ’©
o
5* re j-
"i § o © 3
3 CX5 O o 03 TD
CO 0
<0 0

re
E m
0
00
>N
0
22 CO CO
3
0
LD re re re
0
lo
> 0 0
© ^ re
p re -q ©
c c
>,
c £ c
— c
re c © re re re 0 re
E E E
E ^ 3
© 0
E ll o 0 0 0
re hi co CD C5 CD < CD
c 03
o CO
+J 03
©
<23 *3
00 cr ^r h-
03 o CO 00

3
l— re 03 03

cr ^r o c C
CO 0 c c 03
03 o re re 00
c E E 03
>* 0 0
3 0 0
■o 0
> cr C C 0
3
re 0 0 ©
4-» lD
CO CL CO Q_ Q_
 630 Classroom second language acquisition

o n ■D o 0 ■O
§ s 0 T3 0 o 0 ©> o
0 C
c .5 > 0 +- 0 #f s 0 0 "O o
o o ■O -1= 0
.2 u_,
00 0 0
0 at! !l0 n
03
3 '0 ~ ^0 0 HE
.0
O
3
.>
0
0
*-
it
0 0 c
iE T3
~ O
— C O 8 c
2 ® ai ^ 0
^ o
03 C I!
ffitjO
C 0 o _ 0
■o t!
-o s> c T3 o T3 _0 c 9 3 ° o 9 cog.
fe ®
0 D O -* 0
O 0
0 o o 0 Q_
0 0
0 0 c -Q O to C <=-
0 o
0 .E
0 .
trt
0 ^5 11
0
•> i 0 >_ 0 0 0 02 £ CO
Q- O
vu ° 0 0 "O
§ § E 0 0 '0 c
O 0 0 2 0 -£ q +- 3 3 -o to o
0
0 0
w
>tt®3
b c c
£ E .1 §
© E
-n 0
0 0 ~ 0 -Q
0
C
0
O
3
O C C 0 E
g « C
o -o
0
0 0 > C §-§ "S .<2
_ 3 1 £ o E O o 2s
® § ® 0
Sj>o° 0 -C O - o ligto
3 Q. Q.
o c
£.E
0 Q. iz o .eg
O M_
C O o- B
II
T30 0 -0 I!
^ 5 9-"3 03 ° -p g 0 Q- E ® W
E C
0 0 E § . .. c <0 ° Q. c
■o 9-
0 «b
■O
O _ <n to
Q §
0 O 3 O Ul £ M -o 0 ° o
o) 2 ® Q- 0
C -^3
q.^ "S .9 o o
c ^ O '-P
0 3 0 © -x .E C O
■to ® 0n p S 5
^-£ 0
3 •-P 3 0 JZ 0 0 -g E 0
0 3 ~ °
</)
0
D ^ o O ±3
P m ° 8 c= 0 8|
0 iz
C 0
*£
C
C
O
I
b
c
=
o
o S 8
E 0
c ©
0 0 E
0 0c
ii
0 o
0 "O
2 o> <2
0^ ^ ? - 2
cc C3 E 0 '
3 C
O .E 0 C 3 O 0 o 3 Q. f- £ 3 O) c s i CO JE ^ o § 2 cl

c
c .9 o
0 O 0
■O o 11
■o| 0 +-*
0 0
0 03 S g
*; o O) 03 O p
>, c C C
0 O) p
§5

u
~ 'c w ^2
0 0
0 '/-n
o .2, O 0
0 0 8 §
0 > 0 TD C i ®
| 8 0 0 sz *0 o O O o 5j
b
E
c
© 11
U- 0
n
C
0
0
c
0
_0
Q.'O
0
£
0
<o
-Q
0 C 3 0 C c 0 c ■*-*
03 2 o O O 0 0 ® <s
A ©
Q o a C Q. CO CO O -O w ii

•O
c
0 CO
0 0
0 >
0
~o ■
> 0 0 0 0 0
o 0 E .ra 0 0 1=
c 3 0 T3 ~o !q
Q) a3 "D
0 > -2 0 0 0 0
O 0
E ■D
i— .£ E E E
0
o 0 g a5 0 0 X
o o ■*-'
0. _l _C C
Jj .£

>. 0 ^ 0 ^
0 >>
co > 0
0 co
8'e 0 5-
W (/) ‘0 —1
0
> c Q) O
0 g II II > CO
‘c 0 3
3 C o 'c +
0 — >> 0 0 3 T .
•(-> 0 c 3 © 3 0
O 'co - S -ST
0 0 0 CO
1 c ■o ©
0 © 0 -g ® |
n 3 3 3 > O Q. O
© 0
Q. 0 T3 E
3 X3 T3 "O 'c T— 0 q 3 CO
CO 0
CO < < < 3 CM ~3 f— —) § to ®

0 0
0
0
CO 0 CO
■O Cl ■O
0 0 0
E c c C ■D
3 3 0 c
0 0
0 E 0 0 CL Q_
ll -a 0 0 o
~3 “3 CL
C C _C _c
_J _1 _l
CO LL LL LL
LU ^ .E
LU c LU LU LU

-*-<
0
CO '0
0 co
"o §
03
3 C •o
O c
0 0
0 0 JO 0
Hz C\J c 0
o CO LO
CO
0 03
03
03
03
0
x:
-X
0
03 c T“ CD • •
03 b 0 £
> 0
(/) 0 CL 0 o
T3 0
0 -Q
E 0 0 N r-
3 C c E 3 o cn
■*-> 0 O o O 3
o o -5 03
CO 0 N LU “3 “3 O 0 CD i-
 Formal instruction and second language acquisition 631

It should also be noted that, even though no difference was found between
the orders of instructed and naturalistic learners in most of the studies, there
was evidence in some studies that specific morphemes were performed more
accurately as a result of instruction (see, for example, Perkins and Larsen-
Freeman and Pica, and the discussion in the previous section).
The results obtained by the morpheme studies can be considered, at best,
only weak evidence that formal instruction has no effect on the develop¬
mental route. As we saw in Chapter 3, the methodology of the morpheme
studies has been strongly criticized (Hatch 1978b) on a number of grounds.
Also, the theoretical view of L2 acquisition which underlies these studies is
extremely doubtful. Learners do not acquire the L2 grammar as a set of ‘accu¬
mulated entities’ (Rutherford 1988: 5) but rather work on a number of fea¬
tures simultaneously, gradually sorting out the form-meaning relationships
which they encode, and on the way constructing a number of unique interim
grammars in the manner described in Chapter 4. Finally, as we have already
seen, morphological features such as those investigated in the morpheme
studies can pose substantial learning problems and are often not acquired by
even advanced L2 learners (see Bardovi-Harlig and Bofman 1989). The simil¬
arity in acquisition orders of instructed and naturalistic learners is perhaps
best treated as further evidence of this.
A number of other studies that have focused on the order of acquisition in
specific grammatical sub-systems also suggest that instruction is powerless to
alter the general pattern of development. In Chapter 3 we saw that the func¬
tions of relative pronouns are acquired in a predictable order that reflects
their typological status. Pavesi (1984; 1986) found this ‘universal’ order of
acquisition in both a group of instructed EFL learners in an Italian high
school and a group of waiters learning English naturalistically in Edinburgh,
Scotland. Once again, however, this study found that the instructed group
showed higher levels of acquisition, both in the sense that they had progressed
further along the universal order and in the sense that the errors they mani¬
fested were less ‘primitive’ (for example, they used pronoun copies rather
than noun copies).
More interesting, perhaps, is the research that has examined the effects of
formal instruction on the sequence of acquisition of a number of syntactical
features. Felix and Hahn (1985), in a study based on the same learners studied
by Felix (1981), found that they acquired pronouns in an identical fashion to
naturalistic learners. In this case, the instruction took the form of explana¬
tions and practice of each separate pronoun form in an incremental fashion.
However, the learners did not master the pronouns one at a time but rather
one feature at a time (i.e. gender was acquired first, followed in turn by per¬
son, number, and case). Felix and Hahn argue that the learners made use of
‘instruction-independent learning strategies .
Research on German word order rules has provided the clearest evidence
that formal instruction is unable to affect acquisitional sequences. This work
632 Classroom second language acquisition

is based on studies of naturalistic learners which suggest there is a definite se¬

quence of acquisition of word order rules (see Chapter 3). Pienemann (1984),
in a frequently cited experimental study, showed that instruction in inversion
was successful in the case of a learner who had reached the immediately pre¬
ceding stage, but not in the case of another learner who was at a much earlier
stage. Subsequent longitudinal studies of the classroom acquisition of Ger¬
man by university students in Australia (Pienemann 1986 and 1989) further
demonstrated the immutability of the word order sequence and the failure of
instruction to alter it. My study of 39 adult learners of L2 German in higher
education in Britain (Ellis 1989b) also suggested that instruction is powerless
to alter the sequence of acquisition of these word order rules. This study, as
we have already noted, indicated that the instructed learners appeared to pro¬
gress along the sequence at a much faster rate than naturalistic learners. Two
cross-sectional studies (Daniel 1983 and Westmoreland 1983) also show that
the sequence of acquisition for German word order rules is the same in in¬
structed and naturalistic acquisition.
Research on word order rules in other languages has produced similar res¬
ults. Thus, whereas in Ellis (1984c) I found that many of my learners failed to
acquire inversion in wh- questions in L2 English, probably because they had
not yet developed the necessary processing requisites, Brindley (1991) reports
that instruction that was more carefully tailored to the learners’ level did res¬
ult in substantial acquisitional gains in English question formation. He found
that the eight learners in his study produced considerably more questions af¬
ter the instruction, and that they progressed through the stages in the order
predicted by the Multidimensional Model. He notes, however, that even after
600 hours of instruction (not all of which was directed at questions), the
learners were still using a relatively restricted range of question forms. Even
with instruction the acquisition of complex rules takes time.
There is also the question of what happens if learners are given instruction
in a structure before they are developmentally ready for it. Do they simply fail
to learn it? Or does it have some other effect? A study by Pienemann (1987)
suggests that premature instruction can have a deleterious effect by inducing
avoidance behaviour. He provides evidence to show that three adult class¬
room learners of L2 German tended to avoid using the present perfect, pre¬
ferring modal + verb structures because these gave them a greater chance of
conforming to target-language norms, as the main verb consisted of the
unmodified infinitive. According to Pienemann, the present perfect is ac¬
quired late, as a Stage 5 structure (see Chapter 9). Pienemann suggests that the
avoidance was the result of being forced to produce this structure at an early
stage. Pienemann’s study, based on acquisition sequence, concurs with the
studies referred to earlier, which provide evidence that formal instruction can
inhibit the normal processes of acquisition.
Whereas in the case of the German word order rules, instruction only
works if it follows the developmental sequence, the acquisition of relative
 Formal instruction and second language acquisition 633

pronoun functions appears to be accelerated if learners are taught the more

marked, (i.e. typologically less frequent) functions first. Gass (1982) and Eck-
man, Bell, and Nelson (1988) conducted studies based on the Accessibility
Hierarachy for relativization (see Chapters 3 and 10) and found that instruc¬
tion directed at the object of preposition function (for example, ‘This is the
man who we spoke about’ ) triggered acquisition of the less marked features
(direct object and subject functions), as demonstrated in a written production
task. The reverse, however did not occur to the same extent: for example, stu¬
dents taught the subject function did not show much gain on the object and
object of preposition functions. Subsequent studies by A. Jones (1991b and
1992) replicated these results for Japanese college-level learners of English,
and also showed that the predicted order was evident in oral as well as written
production.2 A study by Zobl (1985) also lends support to the triggering ef¬
fect achieved by teaching marked structures. It provides evidence to suggest
that learners who received instruction directed at the marked use of possess¬
ive pronouns (possessive + human entity, for example, ‘his mother’) im¬
proved in both this feature and in the unmarked use of possessive pronouns
(possessive + inanimate entity, for example, ‘her car’). However learners who
received instruction in the unmarked feature did not show any improvement
in the marked feature and, in one study, not in the unmarked feature either.
These studies indicate that some structures may be hierarchically linked in
such a way that learners are able to generalize knowledge of marked features
to unmarked features but not vice versa.
There is some evidence to suggest that formal instruction directed at fea¬
tures that are not subject to developmental constraints can succeed irrespect¬
ive of the stage the learner has reached. Pienemann (1984) found that learners
at different stages of development who were taught German copula were able
to perform this structure much more accurately. Other studies have also
shown that the accuracy of ‘simple’ structures improves with instruction (see
the previous section). Pienemann suggests that variational features (see Chap¬
ter 9), such as copula, are amenable to instruction.
It is also possible that developmental constraints on formal instruction do
not occur when acquisition is considered in terms of comprehension, as op¬
posed to production. Buczowska and Weist (1991) compared the acquisition
of temporal locatives in English (indicators of past v. present and progressive
v. aspect) by 60 classroom L2 learners, who differed in the total amount of in¬
struction they had received, with that by 60 LI learners, who were divided
into five groups according to age. Comprehension data were collected by
means of a sentence-picture matching test. The study found that the L2 learn¬
ers did not follow the acquisitional route prototypical for the LI learners.
Neither did they follow the pattern of development of untutored L2 learners,
as reported in studies based on learner production (for example, Meisel 1987
and Bhardwaj, Dietrich, and Noyan 1988). It is possible, therefore, that ad¬
ults are capable of comprehending form-meaning distinctions which they
634 Classroom second language acquisition

cannot yet produce and that instruction can help them to achieve this. This is
a point that we will take up later when we consider the effects of different
kinds of formal instruction.
As we noted in Chapter 9, studies which demonstrate the existence of de¬
velopmental patterns are interesting, but without a theory to explain them
they will have limited applicability. This is one of the weaknesses of the
morpheme studies. The Multidimensional Model proposed by Meisel, Clah-
sen, and Pienemann (1981) and subsequently developed by Pienemann and
his associates in Australia, constitutes one the strongest theories of L2 ac¬
quisition currently available (see the account of this theory in Chapter 9).
This theory predicts that instruction will only succeed in teaching a learner a
new developmental structure if the learner is ‘ready’ to acquire it. On the basis
of the theory, Pienemann (1985) advances what he calls the teachability
hypothesis:

The teachability hypothesis predicts that instruction can only promote lan¬
guage acquisition if the interlanguage is close to the point when the struc¬
ture to be taught is acquired in the natural setting (so that sufficient
processing prerequisites are developed) (1985: 37).

Pienemann goes on to discuss the relevance of this hypothesis to syllabus

design, arguing that if grammar instruction directed at a developmental fea¬
ture is to be effective, it must be timed to take place when the learner is ready
for it.
The studies based on linguistic markedness also have a theoretical basis.
Eckman (1977) has put forward the Markedness Differential Hypothesis,
which makes predictions about the areas of difficulty a learner will experience
on the basis of markedness relations between native- and target-language fea¬
tures (see Chapter 8). He claims, for instance that L2 learners will only experi¬
ence difficulty with target-language features that differ from native-language
features if they are more marked. Zobl (1983b) has argued that languages are
enormously complex and cannot be acquired purely on the basis of input.
Learners must possess a projection capacity which enables them to go beyond
the information available to them in the input (see Chapter 10). Both Eckman
and Zobl propose that acquisition along one parameter entails acquisition
along another related parameter. This simplifies the learning task. The im¬
plication for formal instruction is that it may be most effective if it focuses on
those marked grammatical properties which are able to trigger acquisition of
associated unmarked properties. It should be noted, though, that teaching
learners marked grammatical properties does not disturb the ‘natural’ order
of acquisition, but rather hastens progress along it by enabling learners to ac¬
quire a number of features at the same time. Also, as Eckman points out, it is
not possible to conclude on the basis of this theory that instruction will be
most efficient if it focuses on marked properties. Spending a short period of
time teaching each structure in the order in which it occurs on the hierarchy
 Formal instruction and second language acquisition 635

may work out as more effective than spending a longer period teaching a rel¬
atively marked feature.
It is not easy to reconcile the theoretical positions of the Multidimensional
Model and the Projection Hypothesis.3 In particular, it is not clear whether
instruction should seek to follow the natural order of acquisition or should
try to teach the more marked features in the hope that learners will generalize
their new knowledge to implicated unmarked features.
Studies that have investigated the effects of formal instruction on the order
and sequence of acquisition were summarized in Table 14.3. The following
tentative conclusions attempt to reconcile the various findings:

1 Instructed learners manifest the same order of morpheme acquisition as

naturalistic learners (for example, Fathman 1978).
2 Instructed learners also manifest the same order of acquisition of fea¬
tures comprising grammatical sub-systems such as relative pronoun
functions as naturalistic learners (for example, Pavesi 1986).
3 Grammar instruction may prove powerless to alter the natural sequence
of acquisition of developmental structures, as these are manifest in learn¬
er production (for example, Pienemann 1989).
4 Premature instruction may cause learners to avoid using structures and
so may inhibit acquisition (for example, Pienemann 1987).
5 Grammar instruction can be effective in enabling learners to progress
along the natural order more rapidly. One way in which this might be
achieved is by teaching marked features within the sequence (for ex¬
ample, Eckman et al. 1988).
6 Grammatical features that are not subject to developmental constraints
may be amenable to instruction (for example, Pienemann 1984).
7 Formal instruction may help learners to comprehend the meanings of
grammatical structures, even if it does not enable them to use the struc¬
tures in production (for example, Buczowska and Weist 1991).

Again, though, caution needs to be exercised. The notions of ‘acquisition

order’ and ‘sequence’ remain controversial. Also, many of these conclusions
are based on very limited research. There is a clear need to conduct more stud¬
ies to investigate the effects of instruction on comprehension, for instance.
It is should also be noted that the research has been concerned with acquisi¬
tion orders and sequences in relation to implicit knowledge. There is no evid¬
ence that explicit knowledge of grammatical rules is acquired in some fixed
order or sequence. Indeed, it would seem that this is unlikely. Thus, formal in¬
struction directed at such knowledge may not be subject to the kind of pro¬
cessing constraints which Pienemann has identified. If, then, the goal of
grammar teaching is explicit knowledge rather than implicit knowledge, it
may not be necessary to take account of the learner’s stage of development.
The teachability hypothesis, as formulated by Pienemann, therefore, may be
636 Classroom second language acquisition

of relevance only for grammar instruction that has implicit knowledge as its
goal.
It is also not at all clear what practical use teachers can make of the research
which has examined the effects of instruction on the order/sequence of ac¬
quisition. How can they ensure that grammar instruction is timed to match
individual learners’ stages of development? Clearly they would need informa¬
tion about which processing operations learners had already mastered. Piene-
mann, Johnston and Brindley (1988) outline a procedure which teachers can
use to establish this, but the procedure is a complicated one calling for the ac¬
curate completion of an observation form in order to identify individual
learners’ developmental gaps. It is doubtful if most teachers will be able to use
this with sufficient precision. To overcome this problem, and also to facilitate
the analysis of L2 data for research purposes, Pienemann and his co-workers
(see Pienemann 1992) have developed a software computer package called
COALA. However, this package requires the data to be prepared manually
before it can be analysed, and so may not be readily usable by teachers. Even if
a reliable method of diagnosing learners’ stages of development can be found,
there is still the problem of how to cope instructionally with the inevitable
variation in proficiency which exists within even a relatively homogeneous
group of learners. In some teaching contexts (for example, those with access
to computer technology), the necessary individualization might be possible,
but in many, probably most, it would be impossible.
Problems also exist in applying the Projection Hypothesis. So far, the evid¬
ence for this hypothesis is restricted to a few structures. Attempts to extend it
to other areas such as adverb position (White 1991) have not proved success¬
ful. It would seem that projection which is predicted to occur on the basis of
markedness relations identified by linguistic theory does not always occur. In
such cases, researchers tend to query the linguistic theory. It is possible, how¬
ever, that it is the Projection Hypothesis itself that is faulty.

The durability of formal instruction

One'possibility that we need to consider is that even in cases where instruc¬
tion appears to have worked, the beneficial effects may be only temporary.
The increased levels of accuracy that result from instruction may prove to be
impermanent, or the acquisition of a new grammatical feature illusory. As
time passes the effects may gradually atrophy and the learners return to sim¬
ilar levels of performance to those observed before the instruction. If this were
to be the case, the utility of formal instruction would be severely limited.
A number of studies already referred to (see Tables 14.2 and 14.3) indicate
that the effects of grammar instruction may not last. Lightbown et al. (1980)
found that the overall scores of the learners they investigated dropped to a
level approximately halfway between that of the pre-test and the immediate
post-test in a follow-up test administered six months after the instruction.
 Formal instruction and second language acquisition 637

Pienemann (1984) found that the gains which one learner made in the accur¬
ate use of the copula as a result of instruction began to disappear after as little
as one week. White (1991) found that gains in the correct positioning of ad¬
verbs were largely lost five months after the instruction.
However, other studies have found that the effects of instruction are dur¬
able. Harley (1989), for example, retested her subjects three months after the
instruction and found that the learners’ improved ability to use French impar-
fait and passe compose as evident in an immediate post-test, had not only
been maintained but extended even further. White et al. (1991) found that in¬
creased accuracy in the formation of questions evident in the same learners
that White had investigated did not slip back to pre-instruction levels. In fact,
the learners were still improving some six months later.
What explanation can be given for these mixed results? One possibility that
seems very plausible is that advanced by Lightbown (1991a). She notes that
whereas learners had few opportunities for either hearing or using adverbs
once the period of instruction was over, the same was not true for questions.
Questions occurred frequently in the classroom input. Also, the intensive ESL
learners asked many questions themselves, and so were able to obtain con¬
tinuous feedback on their ability to perform them accurately. Lightbown
suggests:

... the findings of these experimental studies can be interpreted as showing

that, when form-focused instruction is introduced in a way which is di¬
vorced from the communicative needs and activities of the students, only
short-term effects are obtained.

In other words, for the effects of instruction to be lasting, learners need sub¬
sequent and possibly continuous access to communication that utilizes the
features that have been taught.
Although this is an attractive explanation, it does not appear to account
fully for the results of all the studies mentioned above. It is reasonable to as¬
sume, for instance, that the structures investigated by Lightbown et al.
(1980)—various -s morphemes, copula ‘be’, and locative prepositions—and
that investigated by Pienemann (1984)—copula ‘be’—are frequent in class¬
room input and that opportunities for producing them were available to the
learners. Yet the effects of instruction directed at these features tended to dis¬
appear. It would seem, then, that other factors are involved.
We can only speculate at what these factors might be. One possibility is the
nature of the linguistic feature itself. Lightbown (1991a) noted that whereas
questions are developmental, adverbs may be variational, in terms of the
theoretical framework used by Pienemann (see the previous section).4 Devel¬
opmental features may be less susceptible to influence by input, but once
acquired—through instruction or through communication—they constitute
stable interlanguage rules. In contrast, the acquisition of variational features
638 Classroom second language acquisition

may be more amenable to input but such forms may continue to be unstable
in the learner’s interlanguage and so easily atrophy.
A second possibility, not incompatible with the one above, relates to the
learner’s perception of a grammatical feature. Features such as copula ‘be’
and -s morphemes are not very salient. Thus, although they occur frequently,
they may not be easily perceived in continuous speech. These features may
also not be seen as very important by learners. Whereas they contribute relat¬
ively little to successful communication, structures such as questions and the
distinction between passe compose and imparfait may be considered more
important for message conveyance. If learners are motivated primarily by
communicative need, then they will probably retain only those features that
they perceive to be important for communication, as suggested by Meisel,
Clahsen, and Pienemann (1981) and by proponents of some functional theor¬
ies of L2 acquisition (see Chapter 9). Only if learners are motivated to acquire
native-speaker norms, as a result perhaps of a desire to become integrated
into the target-language culture, or as a result of an instrumental need to pass
an examination that places a premium on grammatical accuracy, will they re¬
tain features that from a purely communicative point of view are redundant.
According to this view, the durability of instructional effects is closely linked
to the nature of the learner’s motivation. It should be noted that these
explanations are subtly different from the one Lightbown has advanced.
Whereas Lightbown emphasizes opportunity for hearing and using a struc¬
ture, these explanations suggest that such opportunity constitutes a necessary
but not sufficient condition to ensure retention of learnt grammar—the learn¬
er needs both to be able to perceive structures in the input and also requires a
reason for remembering them.
It is clearly essential to establish whether the learning that results from in¬
struction persists, and also what factors determine whether it does or not.
This calls for an experimental research design that includes a delayed as well
as an immediate post-test—a design increasingly favoured by L2 classroom
researchers. To date, few definite conclusions are available but there is suffi¬
cient evidence to show that learners retain at least some of the grammatical
structures they have been taught.

The effects of different types of formal instruction

The research that we have considered so far has attempted to answer the
question ‘Does formal instruction result in the acquisition of the features that
have been taught?’. This research has viewed formal instruction generically as
involving a focus on form and the provision of corrective feedback (see
Krashen and Seliger 1975). In taking this view, researchers have been able to
gain insight into the nature of the complex relationship between instruction
and L2 learning, and also to shed light on the processes of acquisition. From
the teacher’s perspective, however, another question is of equal if not of
 Formal instruction and second language acquisition 639

greater importance—’What kind of formal instruction works best?’ We will

now consider some of the research that has tried to answer this question by
focusing on a number of methodological options available to teachers.5

Focus on forms v. focus on form

An important distinction is that between a focus on forms and a focus on

form (Long 1991). Focus on forms refers to instruction that seeks to isolate
linguistic forms in order to teach and test them one at a time. It is found when
language teaching is based on a structural syllabus. Focus on form involves
‘alternating in some principled way between a focus on meaning and a focus
on form’ (Long 1991). It occurs when teachers follow a task-based syllabus,
but focus learners’ attention on specific linguistic properties in the course of
carrying out communicative activities. Long argues that instruction built
around a focus on forms is counter-productive, while that which allows for a
focus on form results in faster learning and higher levels of proficiency.
A focus on form can be achieved in two principal ways. First, activities can
be devised that require learners to communicate while also focusing their at¬
tention on specific formal properties. Second, teachers can elect to provide
corrective feedback on learners’ errors during the course of communication
activities.
A good example of the first way is to be found in Doughty’s (1991) study.6
This compared the effects of ‘meaning-oriented instruction’ and ‘rule-
oriented instruction’ on the acquisition of relative clauses by 20 intermediate-
level ESL students from different language backgrounds. The materials con¬
sisted of computer-presented reading passages, specially written to contain
examples of clauses where the direct object had been relativized. All the sub¬
jects skimmed the texts first. The meaning-orientated group received support
in the form of lexical and semantic rephrasings and sentence clarification
strategies. The rule-orientated group received instruction in the form of expli¬
cit rule statements and on-screen sentence manipulation. A control group
simply read the text again. The results showed that the meaning-orientated
group and the rule-orientated group both outperformed the control group in
their ability to relativize. The meaning-orientated group, however, also dem¬
onstrated an advantage with regard to comprehension of the content of the
text.
The role of corrective feedback in language acquisition has been extens¬
ively debated (see Schachter 1991) and, as discussed in Chapter 13, there has
been a considerable body of research into the nature of teachers’ corrections
in language classrooms. However, there have been few studies that havein-
vestigated what effect, if any, formal corrections have on language acquisi¬
tion.
An investigation by Tomasello and Herron (1988; 1989) provides evidence
that formal correction can have a real effect on acquisition. They compared
the effects of two kinds of instruction directed at problematic constructions
640 Classroom second language acquisition

that lead to overgeneralization and transfer errors in beginner learners of L2

French.7 In one treatment, the problems were explained and illustrated to the
students. In the other, which Tomasello and Herron referred to as the ‘down
the garden path’ treatment, the typical errors were induced and then cor¬
rected. The results of this study show that leading students down the garden
path was the more effective. Two explanations for the results are offered.
First, Tomasello and Herron suggest that the ‘garden path’ technique encour¬
ages learners to carry out a ‘cognitive comparison’ between their own deviant
utterances and the correct target-language utterances. Second, they suggest
this technique may increase motivation to learn by arousing curiosity regard¬
ing rules and their exceptions.
In Tomasello and Herron’s treatment there was a focus on forms, but it can
be argued that the ‘garden path’ option corresponds to a focus on form—the
kind of grammar instruction that Long (1991) favours—in that the errors
that were being corrected were of the kind that students would make natur¬
ally in communication.
Lightbown and Spada (1990) examined the effects of corrective feedback
in the context of communicative language teaching—exactly what Long had
in mind. This study investigated a number of classrooms which were part of
an intensive communicative ESL programme in Quebec. Lightbown and
Spada found that, although the teaching was mainly communicative in focus,
some of the teachers paid more attention to the students’ formal errors than
others. They found that the learners who did receive error correction
achieved greater accuracy in the production of some structures (for example,
the use of the correct ‘There is ...’ in place of the LI induced error ‘It has ...’)
but not of others (for example, adjectival placement).
It is possible, of course, that a focus-on-forms approach could have
achieved the same results (i.e. the avoidance or elimination of errors such as
‘It has ...’). However, Lightbown (1991a) points out that learners she invest¬
igated in an earlier study (Lightbown 1983; 1987) who had had hours of
practice and drill with ‘there is’ shifted to ‘have’ as soon as they were taught it.
She comments:

In the ‘successful’ intensive programme class, the situations in which the

teacher drew students’ attention to their be/have error were precisely those
situations in which the students knew what they wanted to say and the
teacher’s interventions made clear to them that there was a particular way
to say it (Lightbown 1991b: 209).

These studies—together with the theoretical arguments relating to the

need for negative evidence, which we considered in Chapter 10—suggest that
corrective feedback, when it occurs in response to naturally-occurring errors
or in the context of ongoing efforts to communicate, is helpful to L2 acquisi¬
tion. Other studies, however, have found corrective feedback less effective.
 Formal instruction and second language acquisition 641

Cohen, Larsen-Freeman, and Tarone (1991) refer to two studies in which

most of the students failed to manifest any improvement. Rosenstein (1982)
investigated the effects of correcting learners’ ‘public errors’ (errors that they
were told they were making) and their ‘secret errors’ (errors that they did not
know about). He found that after repeated correction the success rate was
only 50 per cent for public and 20 per cent for secret errors. The other study
suggests the reason for these poor results. Alamari (1982) found that 20 per
cent of the advanced adult learners of L2 Hebrew she studied reported not
paying attention to the teacher’s corrections and only 15 per cent said they
wrote down the correction in their notebooks. In other words, correction of
oral utterances may not be sufficiently attended to by learners.
Carroll, Swain, and Roberge (1992) investigated the effects of corrective
feedback on the acquisition of two complex noun suffixes {-age and -ment) by
intermediate and advanced learners of L2 French. The rules for both suffixes
were explained to two groups of learners, but only the experimental group re¬
ceived feedback when they produced errors during practice sessions. The res¬
ults suggested that the correction helped learners—especially the advanced
learners—to retain the words actually taught, but that it did not help them to
construct morphological generalizations (i.e. the learners in the experimental
group did no better than those in the control group in using the two suffixes to
form nouns from verbs they had not been taught). This study, then, also sug¬
gests that formal correction has a fairly limited effect on acquisition.
There are many possible reasons for these mixed results. One is that cor¬
rective feedback is only likely to result in rule acquisition if it occurs in con¬
junction with natural learning processes—when there is a focus on form
rather than a focus on forms. Another possibility, however, as Schachter
(1991) notes and as the Lightbown and Spada (1990) study suggests, is that
the effectiveness of corrective feedback depends on the particular aspects of
language being corrected. ‘Teacher correction needs to be judicious’ (Cohen,
Larsen-Freeman, and Tarone 1991: 144). It is likely that both the conditions
relating to the provision of teacher correction and the choice of feature to be
corrected affect success.
A focus-on-form approach to language instruction is attractive as it pro¬
vides a way of integrating ‘fluency’ and ‘accuracy’ work but it may be prema¬
ture to reject a focus-on-forms approach. For one thing, Long’s main
objections are based on the difficulty of ensuring that instruction directed at
new linguistic features works, while, as we have already seen, there is plenty
of evidence to suggest that formal instruction can help learners improve the
accuracy of features they have already learnt. The options that we will now
consider all involve a focus on forms.
642 Classroom second language acquisition

Implicit v. explicit instruction

Formal instruction can take the form of an implicit treatment, where learners
are required to induce rules from examples given to them, or an explicit treat¬
ment, where learners are given a rule which they then practise using. This dis¬
tinction underlies the ‘language teaching controversy’ (Diller 1978) of the
1960s and early 1970s, in which the claims of an empiricist approach (such as
the audiolingual method) were pitted against those of a rationalist approach
(such as the cognitive code method).8 As we saw in Chapter 13, a number of
comparative method studies (for example, Smith 1970) sought to establish
which approach was most effective. The results were generally inconclusive,
although the GUME project in Sweden (Von Elek and Oskarsson 1975; cited
in Krashen 1982) did show some advantages for an explicit approach. There
were no overall differences between implicit and explicit methods but the lat¬
ter seemed to work better for adult and female adolescent learners of above
average intelligence.
Other studies have focused on the effects of implicit and explicit instruction
in individual lessons rather than in whole programmes. Seliger (1975) found
that adult ESL learners in the United States retained knowledge of a rule bet¬
ter after it had been presented explicitly. Flowever, Hammerley (1975) found
that some grammatical structures were more amenable to a deductive ap¬
proach, while others were better suited to an inductive approach. This finding
is supported by psychological research on implicit and explicit learning. A
series of studies conducted by Reber (Reber 1976; Reber et al. 1980), for in¬
stance, indicates that explicit instruction works when the material to be learnt
is relatively ‘simple’, but not when it is ‘complex’. The crucial factors are (1)
the number of variables to be learnt and (2) the extent to which the critical
features in the input are salient.
Other work in general psychology also points to another factor that may be
important where the explicit presentation of rules is concerned—whether the
rule is presented in isolation or in conjunction with examples (see Gick and
Holyoak 1983). N. Ellis (1991) compared the effects of three kinds of instruc¬
tion 6n adult university students’ ability to learn the rules of soft mutation in
Welsh. These rules require that initial consonants in Welsh nouns mutate (for
example, /t/-*/d/) in accordance with a complex set of contextual factors.
Learners taught implicitly (i.e. given randomly ordered examples of mutating
and non-mutating nouns in different contexts) showed very uncertain know¬
ledge of the rules of soft mutation. Learners taught explicitly (i.e. given an ac¬
count of the rules for soft mutation) developed a solid knowledge of the rules
but could not always make accurate use of them when asked to judge correct
and incorrect noun forms. Learners given a ‘structured’ treatment (i.e. rules
and examples of how to apply them) did best. They learnt the rules and they
were also successful in using them in judging the grammaticality of sentences.
 Formal instruction and second language acquisition 643

Ellis notes that ‘this group alone knows when novel phrases are ungrammat¬
ical’.
On balance, the available evidence indicates that an explicit presentation of
rules supported by examples is the most effective way of presenting difficult
new material.9 However, the effectiveness of an implicit or explicit instruc¬
tional treatment may depend on the type of linguistic material being learnt
and the characteristics of the individual learner.

Practice v. ‘consciousness-raising’

Traditionally, a focus-on-forms approach has involved giving learners the

opportunity to practise. In Ellis 199Id, I identified a number of features of
language practice from the teacher’s perspective: (1) there is some attempt to
isolate specific grammatical features, (2) learners are required to produce sen¬
tences containing the targeted structure, (3) they must do so repetitively, (4)
they are expected to do so correctly, and (5) they receive corrective feedback.
It follows from these features that learners will be aware that they are enga¬
ging in ‘practice’ and, also that they will be aware of what item or structure
they are trying to learn. I go on to suggest that formal instruction might al¬
ternatively take the form of consciousness raising, which differs from practice
primarily with regard to (2), (3) and (4).10 That is, in consciousness-raising
activities the learners are not expected to produce the target structure, only to
understand it by formulating some kind of cognitive representation of how it
works. Whereas practice is aimed at developing implicit knowledge of the
rule, consciousness raising is directed only at explicit knowledge (i.e. there is
no expectancy that learners will be able to use the rule in communicative out¬
put).
In Chapter 13 we examined a number of studies relating to the role of parti¬
cipation in classroom language learning and saw that the results were very
mixed. Overall, there is no evidence to support the claim—which is central to
those who advocate the need for practice—that participation results in learn¬
ing. In this chapter, we have also reviewed a number of studies that suggest
that the extent to which individual learners practise specific grammatical fea¬
tures is not related to the accuracy with which they subsequently perform
them in communication. In Ellis 1991d, I note that these results are not en¬
couraging for supporters of practice. One possible reason for the failure of
practice is that learners are not yet ready to learn the features they are being
taught (see the earlier discussion concerning the effects of instruction on the
order and sequence of acquisition).
My advocacy of consciousness-raising as an alternative to practice is based
on two main arguments. First, consciousness-raising need not involve pro¬
duction by the learner—at least not on the scale required by practice activi¬
ties. Second, because consciousness raising is directed at explicit rather than
implicit knowledge, it does not run up against the teachability hypothesis.
644 Classroom second language acquisition

There is no evidence to suggest that the notions of order and sequence apply
to explicit knowledge and thus there is no need to ensure that learners are
‘ready’ to learn an explicit rule.
The question that then arises is ‘What is the use of explicit knowledge to the
L2 learner?’ Krashen (1977) provides one answer. What he terms ‘learnt
knowledge’ can be used to monitor and thereby to improve the accuracy of
communicative output. At least two other uses can be identified. As we saw in
Chapter 9, when we considered the role of consciousness in L2 learning,
learners who have explicit knowledge of target-language features may be
more likely to notice these features in natural input. Also, the process of cog¬
nitively comparing what is present in the input with what is the current
interlanguage rule is facilitated if learners have explicit knowledge. In these
ways explicit knowledge may have an indirect effect on the development of
implicit knowledge.
There is as yet no research that has directly investigated my claims in Ellis
1991d, although studies by Fotos and Ellis (1991) and Fotos (1993) provide
indirect support. These showed that Japanese college students can form ac¬
curate representations of the rules for dative alternation, adverb placement
and relative clauses as a result of carrying out consciousness-raising tasks that
require them to construct explicit rules from structured input data. Also, be¬
cause these tasks were constructed in accordance with information-gap prin¬
ciples, the learners were also given opportunities to communicate.
‘Grammar’ became the content that the learners ‘negotiated’ in order to
achieve mutual understanding (see Rulon and McCreary’s (1986) study of
content negotiation referred to in Chapter 11).
In Fotos’ (1993) study, 160 Japanese university EFT learners were asked to
complete a number of consciousness-raising tasks directed at the three gram¬
matical structures mentioned above in small groups. One week after complet¬
ing each task, they were given a listening and a dictation exercise, the texts of
which contained exemplars of the target structures under investigation. After
completing these tasks, they were given the full texts in writing and asked to
underline any ‘special use of English’ which they had noticed. The results
showed that the learners who had undergone the consciousness raising re¬
ported noticing all three structures in the input to a significantly greater ex¬
tent than learners in a control group. Fotos also reports that the gams in
knowledge that resulted from the consciousness-raising tasks were main¬
tained in a follow-up test administered two weeks afterwards. This result
contrasts with the results in Fotos and Ellis (1991), where the subjects did not
receive subsequent exposure in specially designed language activities and,
possibly as a result, did not maintain the immediate gains in knowledge evid¬
ent in the post-test. Fotos’ research suggests that (1) consciousness raising dir¬
ected at specific structures can result in subsequent noticing of these
structures in input, and (2) this noticing may help retention of the structures.
 Formal instruction and second language acquisition 645

Consciousness raising provides a logical way of avoiding many of the ped¬

agogic problems that arise from the teachability hypothesis. It is not surpris¬
ing, therefore, that it has received considerable attention in language
pedagogy (see Rutherford 1988). It remains, however, relatively untested.
Also, as I pointed out in Ellis 199Id, not all learners will be interested in or
capable of inducing explicit representations of grammatical rules.

Interpretation v. practice

When we talk about the ‘comprehension’ of grammatical structures we can

refer to two rather different notions. One sense refers to the idea of learners’
conscious efforts to develop an explicit representation of a rule. A second
sense refers to the learners’ ability to identify the meaning(s) which can be
realized by a particular grammatical structure—to the idea of developing an
understanding of what Widdowson (1978) calls the ‘signification’ of gram¬
matical forms. Whereas the first sense underlies the discussion of conscious¬
ness raising above, the second sense informs the idea of ‘interpretation’ tasks.
Interpretation involves (1) noticing the presence of a specific feature in the
input, and (2) comprehending the meaning of the feature. In effect, it corres¬
ponds to the stage of input-processing referred to as intake in Chapter 9 (i.e.
those features that are noticed in the input and taken into short-term and,
possibly, medium-term memory, but which are not yet part of the learner’s in¬
terlanguage system). One of the assumptions of formal instruction directed at
interpretation is that it is psycholinguistically easier to manipulate the process
of intake than it is to ensure that learners accommodate intake by undertak¬
ing the necessary restructuring of their interlanguage systems. Pienemann
(1985) noted that developmental constraints only apply where learner pro¬
duction in communicative language use is involved. He comments: ‘the input
to the comprehension system does not need to be adjusted to the level of com¬
plexity of the production learning task since there are different types of pro¬
cessing procedures in the two systems’ (1985: 53). Like consciousness-raising
tasks, therefore, interpretation tasks serve as devices for avoiding the instruc¬
tional problems associated with the teachability hypothesis.
Interpretation teaching typically requires learners to display their compre¬
hension of input that has been carefully structured to contain examples of
specific rules or items (Ellis 1993b). It involves, therefore, a comprehension-
based approach to language teaching (Winitz 1981). Asher’s (1977) Total
Physical Response, for example, requires learners to demonstrate their under¬
standing of commands, which have been constructed to teach specific lexical
items or grammatical structures, by acting out their responses. We saw in
Chapter 13 that Asher reports very favourable results for this method in com¬
parison with practice-orientated methods such as audiolingualism, not only
when achievement is measured in listening tests but also in speaking and
discrete-item grammar tests.
646 Classroom second language acquisition

VanPatten and Cadierno (1993) report an interesting study in which they

compared the effects of two instructional treatments, one directed at manip¬
ulating learners’ output to effect change in their developing interlanguage sys¬
tems (i.e. practice), and the other aimed at changing the way the learners
perceived and processed input (i.e. interpretation). Learners of Spanish at
university level who received interpretation training relating to Spanish word
order rules and the use of clitic pronouns performed better in comprehension
tests than a group of similar learners who received production training. This
result is not perhaps surprising, as it can be argued that the comprehension
test favoured the interpretation group. However, this group also performed
at the same level as the practice group in a production task that favoured the
latter.
There are strong theoretical reasons—and some limited empirical evid¬
ence—to give support to a focus-on-forms approach based on interpretation.
As with consciousness raising, the problems of ensuring that the instruction is
made compatible with the way learners learn is achieved by delimiting the
goal of the instruction.

Summary and conclusion

It is probably premature to reach any firm conclusions regarding what type of

formal instruction works best, but the research that has been undertaken
lends support to the following compatible hypotheses:

1 A focus-on-form approach that encourages learners to pay attention to

the formal properties of language in the context of trying to communic¬
ate—either by means of meaning negotiation or by corrective feed¬
back—may facilitate acquisition (for example, Lightbown and Spada
1990).
2 If a focus-on-forms approach is adopted, this is more likely to succeed if:
- rules are presented explicitly and supported by examples (for example,
N. Ellis 1991)
- the instruction is aimed at developing explicit knowledge through con¬
sciousness-raising activities (for example, Fotos 1993).
- the instruction is directed at enabling learners to establish form-meaning
connections during comprehension (for example, VanPatten and
Cadierno 1993).
3 In contrast, there is a growing consensus that instruction directed at pro¬
moting growth of learners’ interlanguage systems (i.e. their implicit
knowledge) through production-training is problematic (for example,
Pienemann 1985).

It must be emphasized, however, that there is a need for much more research
before the more traditional approaches to formal instruction (those based on
the notion of practice) can be dismissed. As was shown in the previous
 Formal instruction and second language acquisition 647

section, there is evidence that these approaches can work for some target fea¬
tures. Also, practice may well serve as one of the ways in which learners can
improve their accuracy over linguistic features they have already acquired.
The main objections to practice as an instructional strategy lie in the prob¬
lems of teaching new structures that are developmental in nature. This sug¬
gests that one solution to the teachability problem is to look out for those
areas of the learner’s interlanguage that are in the process of developing. As
Hyltenstam points out, ‘it is precisely in those areas where the learner exhibits
variation that instruction should help the most’ (1985: 128).
For practice to be of any real benefit, however, it may be necessary to en¬
sure that it takes place ‘under real operating conditions’ (K. Johnson 198 8) by
providing opportunities for learners to produce the target structure in similar
circumstances to those that prevail in normal communication. The difficulty
of contriving communicative tasks where the use of the target structure in
production is essential, however, provides a further reason for exploiting
comprehension-based tasks more fully in formal instruction (see Loschky and
Bley-Vroman 1990 for a very full discussion of communicative grammar
tasks).

Learner-instruction matching
The search for the most efficient type of formal instruction has been motiv¬
ated by the recognition that the process of L2 acquisition manifests certain
structural properties that are universally present in language learners. How¬
ever, as discussed in Chapters 11 and 12, learners also vary considerably in
the way in which they set about trying to learn an L2. This suggests, as Nicho¬
las (1985) has argued, that there is no unique ‘best’ path to ultimate commun¬
icative competence. Rather, learners will differ in the kind of instruction that
they respond to best.
It is possible, therefore, that the optimal type of instruction will be that
which matches the individual learner’s preferred approach to learning.
Learner-instruction matching can take place informally, as teachers make ad
hoc adjustments in the nature of the demands they place on learners in the
context of their day-by-day teaching. Arguably, this is the most effective way
of accommodating individual differences (see Willing 1987). However, it is
also possible to make a conscious attempt to ensure a good fit between the in¬
struction and the learner by matching the learner’s ‘aptitude’ to an appropri¬
ate teaching style. Educational research which has investigated the effects of
such instruction is known as Aptitude-Treatment-Interaction (ATI) (Cron-
bach and Snow 1977).
A good example of ATI research is Pask and Scott’s (1972) study. This in¬
vestigated the ability of native speakers to perform classification tasks. The
subjects were divided into ‘serialists’ and ‘holists’ (a distinction that corres¬
ponds loosely to that between field-independent and field-dependent). The
648 Classroom second language acquisition

learners were then placed into four groups in order to investigate the effects
on learning when the instruction matched and failed to match their preferred
learning style. The results were quite striking. The eight learners in the
matched condition, irrespective of their learning style, all achieved higher
scores than the eight learners in the complementary condition.
It should be noted, however, that not all ATI studies in general education
have produced such clear results. Cronbach and Snow (1977) in their review
of ATI research conclude that overall there is only weak support for matched
instruction. Willing (1987) points out that there are many problems with ATI
research. It is not clear, for instance, what dimension of individual difference
is pivotal where instruction is concerned. Also, given that learners can differ
on a number of factors (language aptitude, learning style, personality, etc.),
matching based on a single factor may be ineffective. ATI research in general
has focused on cognitive factors (in particular learning style) and has tended
to ignore affective factors. Another problem is that many of the studies have
failed to ascertain whether the intended ‘treatment’ was actually manifest in
the instructional practices that occurred, as they lacked a ‘classroom process’
element—the same problem that we saw in the case of the comparative
method studies. These methodological weaknesses also apply to research into
ATI in L2 acquisition.
A number of experimental ATI studies are summarized in Table 14.4.
Ideally, an ATI study needs to be factorial in order to account for two or more
independent variables (see Hatch and Lazaraton 1991: 369ff). In such a de¬
sign at least four groups of learners are needed to allow for both a matched
and a complementary condition for each instructional treatment. However,
relatively few of the studies in Table 14.4 are truly factorial. In many cases,
only two groups were used and the interaction between treatment and indi¬
vidual differences was computed statistically after the event. This is a further
weakness in the studies into L2 acquisition.
Four areas of individual difference have been investigated: (1) IQ, (2) ex¬
pert v. novice learners (3) language aptitude, and (4) field-dependent/holistic
v. fieljj-independent/analytic learning styles. With regard to IQ, there is some
evidence to suggest that learners with high levels of intelligence benefit from a
highly structured instructional method, whereas learners with lower levels
benefit from a less systematic approach (see Carrol and Spearritt 1967),
although other studies cited in Skehan (1989) have produced the opposite
results (for example, Maier and Jacob 1966). At the moment no conclusion is
possible.
Good language learner studies (such as Naiman et al. 1978) have indicated
that experience in learning an L2 is an important factor. Nation and
McLaughlin’s (1986) study compared ‘novices’ (i.e. monolingual and bilin¬
gual learners) and ‘experts’ (i.e. multilingual learners). They found that all the
subjects, irrespective of their experience, benefited from a structured as
opposed to a random presentation of stimuli relating to an artificial language.
 Formal instruction and second language acquisition 649

~fcB
o £ 0
SB 2
"o
• Q- c
o 0) 0
*- , 0 1
p t "§9 0 Q 3 B s
; 0 0) E o E »- -Q 3 U_
* g. ■2 5
T3 •— » "O i ® c -o - © *2 5
) o -C I ^ c
£ CO -C "D 3 0 ■' 1- o c ~ c
> XI O J £ o ^
(D C 3
is??
; | 0 -c O O "O
* ® 2? .
o x=
O
-Q O O 8gl
■O k co c
> E E © 3' ~ o ±;
c 30
> P 0 :
“il
on®
^ 8-2 ra ;U«fc
0 .2 £ it ©
2 3 £ £ : © -O
t it . £9 8 .
0 c 0 c i= c ^ ? o
"S “ D) o “ -E .E o ® ®
0 +- C -D ®
■g ~ ©
= o o ±i 4-0 0
l2i n (n 'C:
© -S 2
E
2 o
g> £
o
°
| 3 © <d
) ffi '-O ■“ • 8
5 g -S
§■“ ■O C 9- -X
_
vT -X
_ - o c o
2 s £ -2 §

W^C
.E j= o w
..
' -n ^ T3 T)
0
r,co £
CD . U xi
!fc « 2
® ra -o 2Bm 0 0
0 0 © 0 O
0 o o
w "O ® 0 ^ t 0 5 C 2 ©
“ o O C (J T3 O .Em©
0 0 P
— o c\T © > 0 0
0 5 'n 0 0©
ox©
c U_ 0
ll-
-|o SS-E I E .= C o 2 E c £
~ 0 8 -2 a cl 0 © P
ill! 5 ■§ 2 O "5 = 2? Poo.
I (0 $ i) CO 0 03
8|£
CO £ O
S|
0 E
I
3 Z n © < Q. © C 03
F x 2 ~ .E Z C 0

0 C >
03-C B
TJ .y
O £ xi
®L*0 0
© xz .E^a
0 o
>^c ~ I O 03 ^ o a 0 ©
l ZZr 0 ■O co 0 0-3 Q. ©

Experimental aptitude-treatment interaction studies

Q. 2 Q. © 0 CL © CL ©
O -Q E CL 3 v I ® 0 ©
Q- cl Q- © 0 © CL O
^ C 0 0 CL 0 0 W © O S 8 0 P?
0 © - 0 0 •O -C >
0
.9 -i E E 03 O > x:
s 03
_ 4= © XI
E - 2
E a) b B
o .2
0
o
> CL
O ,©
t. o 3“, "C3
r
© -P
0 5 © ,E
©
a
CL
o
3
Q"
©
o > £ B
§00 © - -g 3 © "O CL
-
2 CD -C 45 ™ 0
c
8 -g it
E -S 0
CO T3 •o p -o p ; cn_ i «2
.I’ll O 0 0 » ra
■O © 0
c 0
C
©
o 0
0
> & o, O
(Q O 4^ c
O P > CL
0 Q. JD 2 0 0
§|1
o .2 c X
© 0 ! g - ±; £ ”
03 ©
5>
i_
|g
o x:
3 o .2 s ® o 1c 3 P 0 .2
■O £ P .2 w © n 0 3 o
© "O 0
iB | 13 0 0 ^
CO
0
> Q- 0
c
O > o
© ^
! 0 0 < T3 CO -Q .E

0 .
Cl ©
0 T3

©
o
ft
> O
x:
C

c
© 0 -Q

i
■$
0 =
73 -S = -C
^ O
3 0 — © ^
0 CL o
0 © 0 —
3
TJ
f 8 « o
■Q ■O ■° >
™ S,c S- •o T3 P XI
0 03
g> |>“o J
X X 0 o <
Table 14.4:

T3
CL
jo 0
CO
© —r
03 CD
0 _ ,E 0
> Q
O 1 iw
0
D. ©
0
O O) ~a
CO
LU 03
0
0
T3
<
,C O
E E
3 -O CL
< CO 1^ it; Jr

co
•o

tr 0
0 n o
X <
650 Classroom second language acquisition

However, they found a strong interaction effect with the other learning con¬
dition they investigated, explicit v. implicit. Whereas the ‘experts’ performed
equally well with both explicit and implicit instruction, the ‘novices’ were
found to do much better in the explicit condition. In a subsequent study,
Nayak, Hansen, Krueger, and McLaughlin (1987) investigated the inter¬
action between experience and use of learning strategies under two condi¬
tions—memorization and looking for underlying rules. They found that the
‘experts’ used a greater variety of different strategies in the rule-discovery
than in the memory condition, but that no such difference existed for the
‘novices’. In reviewing these and other studies, McLaughlin (1990b: 170)
suggests that the ‘experts’ may have developed ‘greater plasticity in restruc¬
turing their internal representations of the rules governing linguistic input’.
It is also possible, as Eisenstein (1980) suggests, that it is not just the
amount of experience, but also the type of learning experiences that indi¬
vidual learners have had, that influences the kind of instruction they prefer
and benefit from most. Learners whose experience is restricted to a foreign
language classroom where a premium is placed on formal grammar training
may be encouraged to develop high levels of conformity and control, as these
appear to be important for success in such environments (see Genesee 1978).
Learners who have experienced a more communicative mode of instruction,
such as that provided in immersion programmes, may develop somewhat dif¬
ferent types of ability. It is possible, therefore, that there are different kinds of
‘expert’ learners.
The third factor investigated in ATI studies is language aptitude. Wesche’s
(1981) study of students of L2 French enrolled in courses designed for Cana¬
dian public service employees is probably the best ATI study in SLA research
to date. Wesche started with the assumption that ‘learning conditions which
are optimal for one individual may be inappropriate for another’ (1981:125),
and that aptitude tests could be used to identify the special abilities and weak¬
nesses of individual learners. Two types of student were identified. Those in
type A had a high overall score on the MLAT and PLAB tests (see Chapter
11). Those in type B manifested a high level of analytical ability but demon¬
strated problems with phonetic coding and listening. The two treatments
consisted of (1) an audiovisual, inductive approach organized around the pre¬
sentation of linguistic structures sequenced according to order of difficulty
and (2) a more deductive, analytical approach, which taught oral and literacy
skills together and provided explanations of grammatical points and of how
to produce specific sounds. In an initial study, type A students were taught by
approach (1), and type B students by approach (2). The results were encour¬
aging. There were no significant differences in achievement between A and B
students, suggesting that the matched condition led to equal achievement.
Wesche then used a standard ATI design, assigning students of both types
to matched and complementary conditions. The results of an achievement test
showed that students in the matched conditions gained higher scores. Also,
 Formal instruction and second language acquisition 651

when interviewed these students reported greater interest in foreign language

study, more initiative in practising French outside the classroom, and less
anxiety in class. In a partial replication of this study with Japanese college
students, Akagawa (1992) failed to find any interaction between degree of
aptitude (high v. low) and treatment (inductive v. deductive). However,
Akagawa did not look at the effect of different types of aptitude.
The favoured individual difference variable in ATI studies in SLA research
is learning style—in particular, field dependence/independence, which, as we
saw in Chapter 11, has exerted considerable fascination for SLA researchers.
In general, the results have not been very impressive.
Hartnett’s (1985) study of learners studying L2 Spanish at university used a
rather unusual method of determining learning style. She tested the lateral
direction of the subjects’ eye movements when thinking, arguing that right
eye movements indicated the use of the left hemisphere of the brain and,
therefore, analytic thinking, while left eye movements indicated right hemi¬
sphere activity and holistic thinking. The learners were given the choice of
joining an inductive or a deductive programme. She hypothesized that those
choosing the deductive method would be indicative of an analytic learning
style, while those choosing the inductive method would have a holistic style,
which proved to be the case. The study showed that, in general, the students
knew their learning style and were able to opt for the method that suited it.
Furthermore, the students who chose the instructional programme that
matched their learning style outperformed those who did not on cloze and
dictation tests.
Of the studies involving field dependence/independence, only Abraham’s
(1985) has produced a significant interaction effect; field-dependent learners
performed better with an inductive treatment, while the field-independent
ones did better with a deductive treatment. The other three studies (Bacon
1987; Carter 1988; Akagawa 1992) failed to find any significant interaction
between FI/FD and treatment. Given the general doubts that surround this
particular construct (see Chapter 11), further studies are probably not war¬
ranted.
Skehan (1989) reaches three conclusions based on his review of ATI
research:

... first, that the completed condition-seeking research has been some of
the most fascinating in applied linguistics; second, that there are many
studies whose interpretation is not clear for research/design reasons; and
third, that abysmally little research of this sort has actually been done.

While many might disagree with the first of these conclusions, there can be
little controversy about the other two. There is limited evidence to suggest
that some IDs (L2 learning experience and language aptitude) interact with
the instructional approach (explicit/deductive v. implicit/inductive) to affect
learning outcomes. This lends support to the common sense conviction that
652 Classroom second language acquisition

such interactions do occur. As Bialystok (1985) puts it, there needs to be a

‘minimal congruity’ between the learner’s preferred learning strategies and
the type of instruction for L2 acquisition to proceed efficiently. ATI research
to date, however, has been restricted with regard to both the IDs and the in¬
structional conditions that have been investigated. In particular, there is a
need for research to examine how the instructional options discussed in the
previous section affect different kinds of learner.

The role of formal instruction: some theoretical positions

Now that we have reviewed the substantial body of research that has exam¬
ined the effects of formal instruction on L2 acquisition, we can try to evaluate
a number of theoretical positions relating to its role.
One of the main issues in language pedagogy is what Stern (1983) calls ‘the
code-communication dilemma’. There are advocates of what Widdowson
(1983) refers to as ‘pure education ... and its associated permissive pedagogy
of non-intervention’. There are also those who argue that while instruction
may not be necessary for L2 acquisition, it does help learners to acquire more
quickly. Finally, there are a number of scholars who maintain that for some
aspects of language at least, formal instruction is necessary.

The ‘zero option’

The zero option advocates the abandonment of formal instruction. As a result

of early work in L2 acquisition which provided evidence of a ‘natural’ route
of development (see Chapter 3), a number of researchers (for example, Dulay
and Burt 1973; Krashen 1982) and also educationalists (for example, New-
mark 1966; Corder 1976; Terrell 1977; Prabhu 1987) have proposed that
classroom language learning will proceed more effectively if language learn¬
ers are allowed to construct their interlanguages ‘naturally’, in the same way
as they would if they were learning grammar through the process of learning
how to communicate. Prabhu (1987), for instance, has argued:
.. /the development of competence in a second language requires not sys¬
tematization of language inputs or maximization of planned practice, but
rather the creation of conditions in which learners engage in an effort to
cope with communication (1987: 1).
The Communicational Teaching Project in southern India, under Prabhu’s
leadership, sought to demonstrate that ‘form can best be learned when the
learner’s attention is focused on meaning’ (Beretta 1989: 233). It should be
noted, however, that Prabhu does not actually claim that grammar cannot be
learnt through formal instruction, only that learning it through communica¬
tion is more effective.
In contrast, Krashen (1982) does argue that grammatical competence can¬
not be taught. His position, which might be referred to as a ‘non-interface
 Formal instruction and second language acquisition 653

hypothesis’,11 is that ‘learning’ does not become ‘acquisition’ (see Chapter 9).
Formal instruction, therefore, is rejected because it does not contribute to the
development of the kind of implicit knowledge needed for normal commun¬
ication. No matter how much the learner practises, explicit knowledge can¬
not be converted into implicit knowledge. Krashen does accept, however,
that formal instruction can contribute to the learning of explicit knowledge,12
although he sees this as a limited role because only rules that are formally
simple and deal with meanings that are easy to explain can be ‘learned’. Most
rules have to be ‘acquired’. Krashen also claims that explicit knowledge is of
limited value because it can only be used in ‘monitoring’ when the learner is
focused on form and has sufficient time.
The zero position, as advocated by Krashen and Prabhu, entails not only a
rejection of planned intervention by means of the presentation and practice of
different items and rules but also of unplanned intervention in the form of
error correction. Krashen (1972: 74) refers to error correction as a ‘serious
mistake’ and argues that it should be limited to rules that can be ‘learned’. He
claims that it puts students on the defensive and encourages them to avoid
using difficult constructions. Also, it is likely to disrupt the all-important fo¬
cus on communication. However, negative feedback in the form of commun¬
icative responses to learners’ efforts to convey messages—of the kind found
in caretaker and foreigner talk (see Chapter 7)—is permitted. Thus, although
systematic correction is prohibited, incidental feedback is allowed. Beretta
(1989), in research based on classrooms in the Communicational Teaching
Project, has been able to demonstrate that such a distinction is pedagogically
operational.
To what extent does the research reported in the previous sections lend
support to the zero option? Some of it does. The finding that formal instruc¬
tion only works where planned language use is concerned (see studies by
Schumann 1978a and Kadia 1988) bears out Krashen’s claim that formal in¬
struction is only useful for monitoring. Also, studies which show that the
‘natural’ route of acquisition is impervious to formal instruction (for ex¬
ample, Pica 1983; Ellis 1989b) help to confirm the non-interface hypothesis.
However, other findings suggest that the zero position is not tenable. A num¬
ber of studies (for example, Harley 1989; White et al 1991) have shown that
formal instruction helps to improve grammatical accuracy, even in
unplanned language use, and that the gains that learners make can be dur¬
able. Other studies (for example, Pienemann 1984) suggest that formal in¬
struction can result in new knowledge if the structure is variational and, in the
case of developmental features, if the learner has already developed the pre¬
requisite processing operations to acquire it. In short, there is substantial
evidence that formal instruction works not just because it happens to supply
comprehensible input for ‘acquisition’ but because, on at least some occa¬
sions, learners actually learn what they have been taught. It is not surprising
654 Classroom second language acquisition

that the zero option has been seriously challenged by a number of scholars
(for example, Stevick 1980; Sharwood Smith 1981; Faerch 1985).

Instruction as facilitation
The essential claim of the facilitative position is that although formal instruc¬
tion is not necessary to acquire an L2, it helps learning, in particular by speed¬
ing up the process of ‘natural’ acquisition. There are, in fact, several different
versions of the facilitative position. One is the interface hypothesis—the
claim that by practising specific structures learners can ‘control’ them (i.e.
that explicit knowledge gradually becomes implicit). The second is the vari¬
ability hypothesis, according to which instruction can directly affect the
learners ability to perform structures in some kinds of use but not in others.
The third, associated in particular with Pienemann, is the teachability hypo¬
thesis. The fourth is that formal instruction helps to make the internalization
of rules easier in the long term by helping learners to notice them. We can re¬
fer to this as the selective attention hypothesis.

The interface hypothesis

According to the interface hypothesis, instruction facilitates acquisition by
(1) supplying the learner with conscious rules, and (2) providing practice to
enable them to convert this conscious, ‘controlled’ knowledge into ‘auto¬
matic’ knowledge. Sharwood Smith (1981) builds on the work of Bialystok
and McLaughlin (see Chapter 9) in order to develop a full interface model. He
claims that ‘it is quite clear and uncontroversial to say that most spontaneous
performance is attained by dint of practice’ (1981: 166). Stevick’s (1980)
‘Levertov Machine’ also allows for ‘learnt’ knowledge to become ‘acquired’.
He suggests that ‘acquired’ knowledge is developed through communicating
and is stored in tertiary memory. ‘Learnt’ knowledge relates to secondary or
medium-term memory. In certain cases learners can make use of ‘learnt’
knowledge while they are communicating and, thereby, ‘acquire’ it.
The main problem with the interface hypothesis is that it does not give re¬
cognition to the difficulty of altering developmental sequences; learners do
not ‘acquire’ structures they are not ready for, no matter how much they prac¬
tise.

The variability hypothesis

The variability hypothesis differs from the interface hypothesis in one major
respect: it claims that teaching learners new structures will affect their careful
style but not their vernacular style (see Chapter 4 for an account of the styl¬
istic continuum). Thus, its effects will be evident when learners are per¬
forming in planned language use but not in unplanned language use. In Ellis
 Formal instruction and second language acquisition 655

1987c, I claim that the explanation for this lies in the relationship between
different types of classroom interaction and the learner’s variable inter¬
language system:

Participation in the kind of planned discourse that results from teacher-

directed language drills leads to the acquisition of target language norms in
the learner’s careful style — Participation in the more freely-structured
discourse that results from unfocused activities requires the performance of
a greater range of speech acts and induces the negotiation of meanings re¬
quired for the development of the vernacular style (1987c: 191-2).
In addition to the direct effect that formal instruction has on learners’ care¬
ful styles, it can have an indirect effect on their vernacular styles. Tarone
(1983) has suggested that there is movement along the stylistic continuum
over time, while Dickerson (1974, cited in Tarone 1982) claims that the
continued advancement in the careful style may have a ‘pull effect’ on the
vernacular style. In other words, forms that enter a learner’s interlanguage in
the careful style will gradually become available for use in unplanned
discourse.
The variability hypothesis can account for a number of the research find¬
ings. It explains why some researchers found instruction to affect planned but
not unplanned language use. It explains why instruction is powerless to alter
the ‘natural’ route of acquisition, as this is evident only in the learner’s verna¬
cular style. It also explains why instructed learners do better overall than un¬
tutored learners, as the former benefit from the ‘pull effect’ of a more
developed careful style. In one respect, however, the variability hypothesis is
unsatisfactory; it fails to account for why under certain conditions formal in¬
struction can have a direct effect on learners’ unplanned production.

The teachability hypothesis

The third version of the facilitative position—the teachability hypo¬
thesis—has already been considered briefly. It has a considerable history (see
Nickel 1973; Bailey, Madden, and Krashen 1974; Valdman 1978) but by far
the most detailed proposal has come from Pienemann (1985). His main pro¬
posals are:
1 Do not demand a learning process which is impossible at a given stage
(i.e. order of teaching objectives to be in line with stages of acquisition).
2 But do not introduce deviant forms.
3 The general input may contain structures which were not introduced for
production
(1985:63).
The main difficulty arises with (1). Lightbown (1985a) has rightly pointed
out that our knowledge of natural acquisition sequences is still too limited to
make specific recommendations about how they should be related to teaching
656 Classroom second language acquisition

sequences. Pienemann and his fellow researchers have subsequently at¬

tempted to overcome this problem by extending our knowledge of
acquisition sequences, by developing a broad theoretical framework with
predictive power (see the account of the Multidimensional Model in Chapter
9) and by devising means for diagnosing a learner’s stage of acquisition. Nev¬
ertheless, as we noted earlier in this chapter, problems still remain. In particu¬
lar, as Long (1985b) has observed, Pienemann’s proposals rest on the
challengeable assumption that the best way to facilitate natural language de¬
velopment is through the presentation and practice of a series of discrete lin¬
guistic teaching points (i.e. a focus on forms).

The selective attention hypothesis

The fourth version of the facilitative position, the selective attention hypo¬
thesis, is that formal instruction acts as an aid to acquisition, not by actually
bringing about the internalization of new linguistic features, but rather by
providing the learner with ‘hooks, points of access’ (Lightbown 1985a). The
position is reflected in Sharwood Smith’s Pedagogical Grammar Hypothesis
(see Sharwood Smith 1980, as cited in Rutherford and Sharwood Smith
1985):
Instructional strategies which draw the attention of the learner to specific¬
ally structural regularities of the language, as distinct from message con¬
tent, will under certain conditions significantly increase the rate of
acquisition over and above the rate expected from learners acquiring the
language under natural circumstances where attention to form may be
minimal and sporadic (1985: 275).
In other words, instruction does not enable learners to fully acquire what is
taught when it is taught, but prepares the way for its subsequent acquisition.
As Gass puts it, instruction ‘triggers the initial stages in what eventually
results in grammar restructuring’ (1991: 137). Instruction works by helping
learners to pay selective attention to form and form-meaning connections in
the input. It provides learners with tools that help them to recognize those
features in their interlanguages which are in need of modification.
Pedagogical rules constitute one type of tool for facilitating selective atten¬
tion. The question arises as to whether learners are capable of learning such
rules (see the discussion of explicit knowledge in Chapter 9)*^A.,study by
Seliger (1979) suggests that the conscious rules that learners learn as a result
of instruction are often anomalous. Different learners ended up with different
versions of the same pedagogical rule (in this case the difference between ‘a’
and ‘an’). Sorace (1985), in a small-scale study of learners of L2 Italian, also
found that the learners reformulated the rules they had been taught and that,
as a result, the rules were clearly inadequate both as ‘linguistic’ and as ‘ped¬
agogic’ rules. However, Sorace suggests that the ability to actually verbalize a
rule occurs only as the final stage of metalingual development, and thus
 Formal instruction and second language acquisition 657

represents ‘an advanced specialized form of it’.13 It is possible, therefore, that

the conscious rules learners develop are sufficient, even at an early stage, to act
as ‘acquisition facilitators’ by focusing learners’ attention on ‘critical attri¬
butes of the real language concept that must be induced’ (Seliger 1979: 368).
A second type of tool is the interpretation task—a device for helping learn¬
ers to notice and comprehend features in the input. We considered a number
of studies that suggest that such a tool is effective in promoting intake (for ex¬
ample, VanPatten and Cadierno 1993) and, ultimately, the restructuring re¬
quired for interlanguage development (for example, Doughty 1991).

Overall, the research we have considered in this chapter lends considerable

support to the facilitative position. We have seen that instruction appears to
result in faster learning and higher levels of proficiency (Long 1983c). There is
research to support each version of the position, but the selective attention
hypothesis is probably the most tenable. Practice does work, but not always.
In particular, practice may often fail to bring about any development in the
learner’s vernacular style. Teaching learners structures they are ready for,
however, can result in ‘full’ acquisition and can cause learners to use the
structures in a wide range of linguistic contexts, but there are practical prob¬
lems in achieving this. Facilitating selective attention by devising instruc¬
tional activities that equip learners with conscious rules, or that help them
interpret the meanings of specific forms in the input, is both psycholinguist-
ically feasible and possible in practical terms. Finally, we should note that
formal instruction may prove most facilitative when it matches the learners’
own preferred learning styles.

The necessity for instruction

There is little, if any, support for the claim that classroom learners must have
formal instruction in order to learn the L2. Despite reservations regarding
‘the permissive pedagogy of non-intervention’, there is general recognition
that much of the language learning that takes place in the classroom takes
place ‘naturally’, as a result of learners processing input to which they are ex¬
posed. It is possible, however, that there are certain linguistic properties that
cannot be acquired by L2 learners (especially adults) unless they receive in¬
struction in them.
One occasion on which instruction may be necessary is when the learner is
in danger of constructing an over-inclusive grammar. We saw in Chapter 10
that principles such as the Subset Principle may not operate effectively in L2
acquisition, with the result that certain types of ‘problematic overgeneraliza¬
tion’ occur. According to White (1991b), this is likely to arise when the learn¬
er’s LI is more general than the L2, as shown in Figure 14.2. Such a situation
arises when francophone learners of L2 English attempt to insert an adverb
between the verb and the direct object (for example, "‘John drank yesterday
some coffee). White argues that this type of error cannot be
658 Classroom second language acquisition

eliminated purely on the basis of the positive evidence supplied by communic¬

ative input, because the learner could never be sure that such sentences were
not possible. In such cases, negative evidence in the form of a grammar lesson
or corrective feedback is required.14

Figure 14.2: Cases where LI is more general than the L2 (from White 1991b: 195)

It has also been suggested that even ‘benign overgeneralizations’ (i.e. errors
that can be eliminated on the basis of positive evidence) may require instru¬
ction. Rutherford (1989) suggests that learners may fail to expunge such errors
on the basis of positive evidence because they have come to understand that
language tolerates synonymy. Thus, noticing ‘went’ in the input may not be
sufficient to eliminate ‘goed’, if the learner operates with the hypothesis that
both forms are possible. It follows that learners need to have the fact that
‘goed’ and ‘went’ are not acceptable synonyms brought to their conscious at¬
tention.
In effect, these arguments bear upon the results of the research we con¬
sidered in Chapter 13, which indicated that even under favourable conditions
classroom learners fail to develop full L2 linguistic competence simply by
communicating. It should be noted, however, that it does not follow that for¬
mal instruction is the answer. It is possible that many adult learners will fail to
develop high levels of grammatical competence no matter what the instruc¬
tional conditions. d’Anglejan, Painchaud, and Renaud (1986) conclude their
study of the effects of intensive mixed instruction (teaching that combined
form-focused work with more communicative activities) on adult immigrant
learners of L2 French in Canada as follows:

It is both disappointing yet challenging to discover that after 900 hours of

formal instruction, the vast majority of the subjects have attained profi¬
ciency levels which at best can be described as minimal (1986: 199).

In other words, there may be limits to what is achievable through classroom

learning for the simple reason that there are limits regarding what most learn¬
ers are capable of achieving under any conditions. Certainly, White’s (1991)
finding that the beneficial effects of instruction on adverb placement wore off
over time would lead to such a conclusion, as would Brindley’s (1991) finding
that even after intensive instruction the learners in his study showed relatively
little development in interrogative forms. It is possible, however, that formal
instruction may produce better results with learners who possess better
 Formal instruction and second language acquisition 659

literacy and metalinguistic skills (such as university learners) than the learners
in these studies, but this still remains to be demonstrated.

Conclusion

This chapter has reviewed research and theoretical positions relating to the
role of formal instruction on L2 acquisition. As we saw in Chapter 13, there
are major weaknesses in the second language classroom research. These ap¬
ply just as much to studies of formal instruction as to other kinds of classroom
research. It is important, therefore, to exercise caution in reaching conclu¬
sions. The following, however, are compatible with the research findings to
date:

1 Benefits of formal instruction

The case for formal instruction is strengthening and the case for the zero op¬
tion is weakening. Formal instruction results in increased accuracy and accel¬
erates progress through developmental sequences. Also, its effects are, at least
in some cases, durable. Formal instruction is best seen as facilitating natural
language development rather than offering an alternative mode of learning.

2 Constraints on formal instruction

There are constraints on whether formal instruction works and, if it does, on

whether the effects are lasting. Factors such as the degree of markedness,
form-function transparency, and the nature of the processing operations
involved determine how difficult different structures are to teach. Also, the
learner’s stage of development affects whether the instruction is successful.
Formal instruction combined with opportunities to experience the structures
in communication appears to produce the best results. For some structures
(for example, those where ‘problematic overgeneralizations’ occur), formal
instruction may be necessary to ensure the target structure is learnt.

3 Types of formal instruction

It is not yet clear which kind of instruction works best but there is evidence to
suggest that focusing learners’ attention on forms, and the meanings they
realize in the context of communicative activities, results in successful learn¬
ing. There may also be a case for consciousness raising directed at helping
learners to formulate explicit knowledge. Although practice activities dir¬
ected at language production may help learners achieve greater control over
structures they have already partially acquired, they may be less successful in
enabling learners to acquire ‘new’ developmental features.
660 Classroom second language acquisition

4 Factors affecting formal instruction

Although it may be possible to establish the general characteristics of effective

formal instruction, allowance will probably also have to be made for vari¬
ations in learners’ learning style, although it is not yet clear what instructional
and learner factors need to be taken into account to ensure an effective match¬
ing.
Finally, it is necessary to acknowledge the impossibility of teaching the
total grammar of a language (see Krashen 1982). It follows that in order to
achieve full competence, learners must rely primarily on positive input. The
issue is, then, not whether instruction should provide access to language as
used in the communicative exchange of meaning, but whether it should also
seek to draw learners’ attention to specific linguistic properties. The answer is
‘yes’; this chapter has tried to show both why this needs to be done and also
how.

Notes:

1 Felix and Weigl (1991) further suggest that formal instruction may result
in factors that ‘lead to the total elimination of any kind of UG control’. The
results of their study gave a clear ‘no’ to the question of whether the sub¬
jects had access to Universal Grammar.
2 Jones (1992) queries whether the order found in the oral production task
can be adequately explained by the Accessibility Hierarchy (AH). He sug¬
gests that although the distinction between DO and OP is clear to gram¬
marians, his learners did not see it the same way. Their performance on the
oral task indicated that they treated verbs with prepositional phrases (for
example, ‘working on the floor’) as if they were VP + NP constructions,
where the verb consisted of a two-word chunk ‘working on’. This bears out
the conclusion reached in Chapter 10, namely that the acquisition of relat¬
ive pronoun functions is best explained in terms of processing difficulty
rather than the AH.
3 ‘Sequence’ theories and ‘projection’ theories are not necessarily contradict¬
ory. First, projection theories do not claim that input rich in marked fea¬
tures enables learners to learn ‘out of sequence’, but only that it will enable
them to acquire a cluster of features simultaneously. Second, as Jones
(1991b) has suggested, projection may only be applicable to the re¬
activation of items that have been learnt earlier but which have become
dormant. It is noticeable that none of the projection studies have examined
the effects of instruction on completely new grammatical structures.
4 Lightbown has subsequently reversed her view that adverbs may be vari¬
ational (personal correspondence). She now thinks they are probably
developmental.
 Formal instruction and second language acquisition 661

5 It is not intended to suggest that these ‘options’ constitute ‘approaches’.

Any particular lesson may be made up by selecting from the range of op¬
tions available.
6 Another example of how a focus on forms can be achieved is Nobuyoshi
and Ellis’ (1993) study (see Chapter 7).
7 Tomasello and Herron’s studies have been criticized by Beck and Eubank
(1991) on both theoretical and methodological grounds. Tomasello and
Herron (1991) have defended themselves against these criticisms.
8 The effectiveness of an explicit instructional treatment is also demon¬
strated by Doughty’s (1991) study, referred to in the previous section of
this chapter. Doughty’s rule-orientated group did just as well as her
meaning-orientated group when it came to learning the target structure
(that is, relativization).
9 The terms ‘explicit’ and ‘implicit’ are not to be equated with the terms
‘analytic’ and ‘experiential’ (see Chapter 13). The term ‘formal instruc¬
tion’ can be equated with ‘analytic’ instruction; such instruction can be
either ‘explicit’ or ‘implicit’. It should be noted, however, that, rather
confusingly, the term ‘implicit’ is sometimes used in connection with
communicative language teaching (i.e. ‘experiential’ instruction) by other
authors.
10 The term ‘consciousness raising’ has been widely used in SLA research by
researchers interested in the role of formal instruction. It is frequently
used to refer to any attempt to focus the learner’s attention on a specific
target structure, i.e. as a synonym for formal instruction (see Sharwood
Smith’s (1981) use of the term). In this sense, language practice activities
also involve consciousness raising. In a later article, Sharwood Smith
(1991) suggests a better term might be ‘input enhancement’. He argues
that we can only know what we do with input, not what effect our at¬
tempts have on the learner (i.e. whether the input actually raises con¬
sciousness). However, the term ‘consciousness raising’ is preferred in this
chapter in acknowledgement of its wide currency. It is used with the nar¬
row definition given in Ellis (1991d).
11 Krashen suggests that his Input Hypothesis constitutes a ‘weaker-
interface position’ on the grounds that learner output produced with the
help of ‘learned’ knowledge can serve as comprehensible input for the
purposes of ‘acquisition’. However, in so far as Krashen argues that
‘learned’ knowledge does not convert directly into ‘acquired’ knowledge,
his position is tantamount to a strong non-interface position.
12 Krashen (1982) also recognizes a second use of formal instruction—it can
serve as subject matter for ‘language appreciation lessons’.
13 Sorace (1985: 244) identifies three stages in the development of conscious
metalingual knowledge: subjects are (1) unable to identify errors in cor¬
rect sentences, (2) able to identify and correct errors but not to state the
grammatical rules concerned, and (3) are able to identify and correct
errors, and also to state the grammatical rules.
662 Classroom second language acquisition

14 It should be noted that White does not argue that formal instruction
should be limited to ‘problematic overgeneralizations’. She also argues
that it can have a valuable facilitative effect when ‘benign errors’ occur.

Further Reading
There are a number of general surveys of the literature on the role of formal
instruction in L2 acquisition, the most recent of which are:
M. Long, ‘Instructed interlanguage development’ in L. Beebe (ed.), Issues in
Second Language Acquisition (Newbury House, 1988).
B. Harley, ‘Effects of instruction on SLA: issues and evidence.’ Annual Re¬
view of Applied Linguistics (1988) 9: 165-78.
R. Ellis, Instructed Second Language Acquisition, Chapter 6 (Black-
well, 1990).
D. Larsen-Freeman and M. Long (eds.), An Introduction to Second Language
Research, Chapter 8 (Longman, 1991).

Perhaps the best way into what is now a very extensive literature is to read a
selection of articles reporting empirical studies dealing with the various issues
discussed in this chapter. The following are recommended as representative
of the kind of research that has been carried out:

1 The effects of instruction on general language proficiency

S. Krashen, C. Jones, S. Zelinski, and C. Usprich, ‘How important is instruc¬

tion.’ English Language Teaching Journal (1978) 32: 257-61.

2 The effects of instruction on production accuracy

P. Lightbown, ‘Exploring relationships between developmental and instruc¬

tional sequences’ in H. Seliger and M. Long (eds.), Classroom Oriented Re¬
search in Second Language Acquisition (Newbury Housel983).
*

3 The effects of instruction on the order I sequence of acquisition

T. Pica, ‘Adult acquisition of English as a second language under different

conditions of exposure.’ Language Learning (1983) 33: 465-97.
M. Pienemann, ‘Psychological constraints on the teachability of languages.’
Studies in Second Language Acquisition (1984) 6: 186-214.

4 Instruction and‘projection’

F. Eckman, L. Bell, and D. Nelson, ‘On the generalization of relative clause

instruction in the acquisition of English as a second language.’ Applied Lin¬
guistics (1988) 9: 1-20.
 Formal instruction and second language acquisition 663

5 The durability of instruction

L. White et al., ‘Input enhancement and L2 question formation.’ Applied Lin¬

guistics (1991) 12: 416-32.

6 Comprehension-based formal instruction

C. Doughty, ‘Second language instruction does make a difference: Evidence

from an empirical study of SL relativization.’ Studies in Second Language
Acquisition (1991) 13: 431-70.

7 Learner-instruction matching

M. Wesche, ‘Language aptitude measures in streaming, matching students

with methods and diagnosis of learning problems’ in K. Diller (ed.), Indi¬
vidual Differences and Universals in Language Learning Aptitude (Newbury
House 1981).
For more theoretical discussions of the role of formal instruction, the fol¬
lowing provide a variety of perspectives:
S. Krashen, Principles and Practice in Second Language Acquisition (Perga-
mon 1982).
M. Sharwood Smith, ‘Consciousness-raising and the second language learn¬
er.’ Applied Linguistics (1981) 11:159-68.
M. Pienemann, ‘Learnability and syllabus construction’ in M. Pienemann
and K. Hyltenstam, (eds.), Modelling and Assessing Second Language Ac¬
quisition (Newbury House 1985).
R. Ellis, ‘Contextual variability in second language acquisition and the relev¬
ancy of language teaching’ in R. Ellis (ed.), Second Language Acquisition in
Context (Prentice Hall International 1987).
M. Long, ‘Focus on form: a design feature in language teaching methodology’
in K. de Bot, D. Coste, R. Ginsberg, and C. Kramsch (eds.), Foreign Language
Learning Research in Cross-cultural Perspective (John Benjamins 1991).
V

*
PART SEVEN

Conclusion
Introduction

In this final section of the book it would be pleasing to demonstrate how all
the separate lines of enquiry which we have pursued in the earlier sections
come together to form a single, well-defined picture. However, this is not pos¬
sible. The object of our enquiry—second language (L2) acquisition—is best
seen as a complex, multi-faceted phenomenon—more like a many-sided
prism than a neat picture with clearly identifiable objects. The images that the
prism presents vary in accordance with the angle from which it is viewed and
the light directed at it, with the result that, while they are in some way inter¬
related, they also afford different perspectives of the same entity. At this stage
in second language acquisition research, it may be possible, as Long (1990c:
656ff) contends, to provide a list of ‘well-established findings about learners,
environments, and interlanguages’—indeed, this has been one of the goals of
the preceding chapters—but it is not yet possible to arrive at a single, compre¬
hensive theory to explain them. Therefore, none will be attempted.
Instead, we will consider the increasing tendency in SLA research to turn in
on itself—to examine its own navel, so to speak—by addressing a number of
epistemological issues of considerable importance to the whole enterprise.
These issues concern the kind of data used in SLA research, theory construc¬
tion, and the application of the findings of SLA research to other fields.
'

<
15 Data, theory, and applications in
second language acquisition research

Introduction

In this chapter, we will address three questions:

1 What kinds of data are required to investigate L2 acquisition?

2 How should theory-building proceed in SLA research?
3 Can the results of SLA research be applied to other fields, in particular,
language pedagogy,1 and, if so, how?

These questions are, of course, interrelated. For example, the kind of data
we collect to investigate L2 acquisition is likely to reflect our theoretical
views. The applicability of SLA research to language pedagogy will depend
on both the perceived relevance of the research (which, in turn, is likely to de¬
pend on the kind of data used), and the nature of the theory advanced to ex¬
plain the results obtained. The nature of these interrelationships will become
more apparent in the following sections.
The aim of this chapter is the same as preceding chapters: to provide an ac¬
count of differing positions as objectively as possible. However, personal
views always encroach on objectivity and in this chapter perhaps more so
than the others. I have followed the procedure of describing alternative posi¬
tions before offering my own views.
The issues discussed in this chapter relate to issues of wide epistemological
concern. However, it is not my intention to embark on a full examination of
them, although there is undoubtedly a need for such as examination. I have
set myself a narrower goal; one that accords with the general aim of this book
(i.e. to provide a balanced and, as far as possible, objective account of work
undertaken in SLA). I shall seek only to identify the issues relating to data,
theory, and pedagogical applications as these have been reflected in the SLA
literature to date.

The choice of data in second language acquisition research

We have come up against the kinds of problems that SLA researchers face
with regard to data at various points in this book: in Chapter 2, where we
considered the data needed to examine learners’ errors; in Chapter 4, where
we noted that the descriptive ‘facts’ of learner-language vary according to the
 670 Conclusion

‘task’ used to collect data; in Chapter 5, where we considered the merits of

discourse completion questionnaires for investigating pragmatic features of
learner language; in Chapter 10, where we examined some of the problems
attendant on using data from grammaticality judgement tests; and in Chap¬
ters 12 and 13, where we saw the importance attached to the use of self-report
data in the study of learner differences and learning strategies, but also noted
the doubts that exist regarding the validity and reliability of such data. As in
other social sciences, there is no agreement about what constitutes the ‘right
stuff’ in SLA research.
We can distinguish three broad types of data in SLA research. First, there is
what might be called ‘language use data’—data that reflect learners’ attempts
to use the L2 in either comprehension or production.2 This type of data can be
further distinguished according to whether it constitutes some form of ‘nat¬
ural language use’ (i.e. it occurs in the course of using the L2 in the kind of
communication that learners engage in when they are not being studied) or
‘elicited language use’, where, following Corder (1976), we can identify two
further categories: clinical elicitation (inducing the learner to produce data of
any sort) and experimental elicitation (inducing the learner to produce data
relating to the specific feature(s) in which the researcher is interested). The se¬
cond major data type consists of learners’ metalingual judgements, of the
kind elicited by a grammaticality judgement test. The third type is self-report
data, obtained through questionnaires, interviews and think-aloud tasks.
These various types are shown in Figure 15.1.
natural

language
use
(comprehension clinical
and
production)

elicited
data metalingual
types* judgements

experimental

self-report

Figure 15.1: The data types used in second language acquisition research
Different kinds of SLA research have tended to favour different data types.
Thus, the early descriptive work that led to claims about ‘order’ and ‘se¬
quence’ showed a clear preference for language use data—natural language
use in the early case studies, and clinical and experimental elicitation in the
cross-sectional studies (for example, the morpheme studies). The study of in¬
terlanguage pragmatics, however, has relied extensively on experimentally
 Data, theory, and applications 671

elicited data, using, for example, Discourse Completion Questionnaires, al¬

though some work has also been based on natural language use. In contrast,
research based on ‘strong’ theories, such as Universal Grammar and the Com¬
petition Model, has relied heavily on learners’ metalingual judgements. Re¬
search into individual differences has found language use data of limited
value, and has turned to various kinds of self-report data. This, then, is the
first, obvious point to be made; the type of data used closely reflects the re¬
search goals. We will now consider the pros and cons of each data type.

Natural language use

The main strength of natural language use is that it provides information

about what learners actually do with the L2 when they try to use their L2
knowledge in communication (i.e. the data can be considered ‘authentic’).
For example, the importance of formulaic speech in L2 production is only
evident in data reflecting natural language use. This kind of data shows the
linguistic and pragmatic features of the L2 that learners typically employ in
different situations.3 It can be used to investigate to what extent and in what
ways learners achieve procedural control over their knowledge. It can shed
light on the socio-affective factors that influence language performance. In
particular, it affords insight into the kinds of strategies that learners employ
in constructing and using their interlanguage. In Labovian terms, natural lan¬
guage use shows us the learner’s vernacular style, which some researchers
consider to be ‘basic’ in the sense that it reflects the most normal way know¬
ledge of language is utilized.
There are also several well-attested disadvantages of natural language use.
The most obvious are that it is time-consuming and difficult to collect, and
may not provide information relating to the specific features that are the tar¬
get of the study, or not in sufficient amounts to make a quantitative analysis
possible. It is no accident that early studies focused on negatives and
interrogatives, as these structures occur frequently in natural language use.
Other disadvantages also exist. As Beebe and Cummings (forthcoming)
comment in a discussion of natural speech act data:

... many studies of natural speech have not given us scientifically collected
speech samples that represent the speech of any identifiable group of speak¬
ers. They do not give us situational control, despite the fact that situation is
known to be one of the most influential variables in speech act perform¬
ance.

The significance of this criticism is that it comes from researchers, who by

virtue of their chosen area of enquiry (interlanguage pragmatics) are disposed
to look favourably on natural speech data. However, Beebe and Cummings
find this type of data ‘unsystematic’. In effect, they are challenging the value
672 Conclusion

of natural language use and asserting the need for elicited language use so that
variables that influence performance can be investigated systematically.

Elicited language use

The elicitation of language use provides researchers like Beebe and Cum¬
mings with the ‘systematic data’ they feel are necessary. Clinical elicitation
has been accomplished by means of various kinds of spoken and written
tasks. Data from written compositions, for example, have been used extens¬
ively in error analyses (see Chapter 2). Information-gap tasks of various kinds
have been widely used to obtain samples of relatively spontaneous speech (for
example, by Light bo wn 1983; Ellis 1989b). Other devices that have figured in
SLA research are oral interviews and role plays. Such instruments provide the
researcher with ample data and also with greater ‘situational control’ over the
data. However, the status of the data is uncertain, as we cannot know for sure
what ‘style’ the data represent. Is the language produced in an information-
gap task a true reflection of the learner’s vernacular style? Also, one of the
problems of natural language use—inadequate information relating to spe¬
cific language features—may not be overcome.
It is the need for data on specific linguistic or pragmatic properties that pro¬
vides the principal justification for experimental elicitation. Larsen-Freeman
and Long (1991) list the various instruments that have been used to elicit ex¬
perimental L2 data: reading aloud (subjects are asked to read words,
sentences, or passages out loud), structured exercises (for example, trans¬
formation exercises, fill-in-the-blanks, and sentence combining), completion
tasks (involving either discrete sentences or whole passages), imitation tasks
(where learners are asked to imitate sentences read to them), translation,
guided composition, and story-retelling. All these devices serve, like tests,4 to
provide the researcher with samples of learner language containing copious
exemplars of the particular property that is the target of the investigation.
The key issue surrounding these experimental elicitation devices is the ex¬
tent to which they supply valid data. We will examine this issue with reference
to one device—the Discourse Completion Questionnaire—which has gained
popularity in the study of interlanguage pragmatics. This device, which we
discussed and illustrated in Chapter 5, provides learners with a situation de¬
signed to elicit a specific illocutionary act such as a request or apology, and
asks them to write down what they would say. The usual way to establish the
validity of the data collected by this instrument is by establishing to what ex¬
tent they display the same features as natural speech data. Thus, Beebe and
Cummings (forthcoming) compared refusals elicited by a Discourse Comple¬
tion Questionnaire with those occurring more or less naturally in telephone
conversations.5 The results of their study suggest that there are both sim¬
ilarities and differences. They found that the elicited data produced shorter
 Data, theory, and applications 673

responses, fewer repetitions and turns, and, in particular, less remedial work
(attempts to mitigate the force of a refusal). But they also noted that the eli¬
cited data provided an accurate picture of the different semantic formulas
used to perform the illocutionary act. In the eyes of Beebe and Cummings this
‘validated’ the data. We should note, however, that concurrent validation of
this kind requires the researcher to collect samples of natural language use in
addition to the experimentally elicited data, which, often enough, has not
been done. We might also note that not all researchers would accept that eli¬
cited data is ‘valid’, even if it can be shown to match natural language use.

Metalingual judgements

This is precisely the position adopted by some researchers who make use of
learners’ metalingual judgements. Asking learners to judge the grammat-
icality of sentences (as in much UG-inspired research) or asking them to
identify which noun functions as the agent or subject in a sentence (as in re¬
search based on the Competition Model) taps into their metalingual know¬
ledge. One justification for such tasks is that they provide evidence about
what learners know as opposed to what they do, about competence rather
than performance. Such a claim is probably not warranted, however. Meta¬
lingual judgements involve ‘performance’ just as much as natural language
use, albeit of a different kind. There is no direct window into competence.
Also, conversely, data taken from natural language can provide evidence of
learners’ competence, not just of their performance. In short, all data, from
whatever source, can be used to investigate both competence and perform¬
ance.6 Metalingual judgement data do not have a privileged status, as is now
generally recognized.
The great advantage of metalingual judgement tasks is that they enable re¬
searchers to investigate linguistic properties that would be otherwise very dif¬
ficult (and perhaps impossible) to investigate. Language use data—natural or
elicited—provide only information about the language that learners produce
or comprehend in particular situations. They tell us nothing about the lan¬
guage learners do not use. Thus, we do not know whether the fact that learn¬
ers fail to produce sentences such as:

*What did Randy wonder whether his brother would win?

is just because they have not got round to doing so or because their in¬
terlanguage grammars do not permit them to do so. Metalingual perform¬
ance data can shed light on this issue.
There are obvious disadvantages, however (see Chapter 10). Birdsong
(1989) has written at length on the ‘logical problem of the data’ collected
by means of metalingual judgement tasks. The pitfalls include the difficulty
of obtaining reliable judgements from illiterate bilinguals, response biases,
the problem of determining exactly what it is that learners judge when they
674 Conclusion

evaluate sentences, and the variability that arises from different learners’
‘skill’ in performing such tasks. Birdsong suggests ways in which these pitfalls
can be overcome. Like others before him (for example, Arthur 1980; Chau-
dron 1983a), he also suggests that metalingual performance data take on
validity when the results they provide converge with those obtained from
other data sources, in particular language use. In fact, though, despite Chau-
dron’s claim that ‘metalinguistic judgements in NSs and NNSs tend to be val¬
idated by other measures of performance’ (1983a: 371), there is plenty of
evidence that they do not (see, for example, the studies of variability in learner
language discussed in Chapter 4).

Self-report

The third major type of data in SLA research is self-report data. These have
proved invaluable in exploring individual differences in learners and identi¬
fying the various learning strategies they employ (see Chapters 11 and 12).
Self-report data have been obtained in various ways: by means of written
questionnaires such as those used by Gardner and his associates to measure
learners’ attitudes and motivation (see Gardner 1985), by means of oral inter¬
views of the kind used by Wenden to identify learners’ beliefs about language
learning (see Wenden 1987a), and by think-aloud tasks, performed either by
individual learners or by pairs of learners and used to identify the nature of
the problem-solving strategies learners employ in various language tasks (see
the papers in Faerch and Kasper 1987). Self-report data has proved invaluable
in uncovering some of the affective and cognitive factors involved in L2 learn¬
ing, factors that are not readily observable in language behaviour.
Again, though, there are problems. One concerns the extent to which learn¬
ers are sufficiently aware of their affective states and cognitive processes to re¬
port on them. This is a problem of validity, but not a serious one, perhaps, as
we can simply acknowledge that self-report data are limited in this respect
and, therefore, like any other type of data, incomplete. Another problem is
that ‘subject’s reports may derive more from what they think they should
have been doing than what from what they actually were doing’ (McLaughlin
1990c: 629). This is a problem of reliability, which might be overcome by
comparing data collected from the same subjects using the same instrument
on two separate occasions and establishing re-test reliability—a procedure
which, regrettably, has been little used in SLA research.

The question of validity

Underlying this discussion of the pros and cons of the different types of data is
the central problem of validity. What constitutes valid L2 data? One answer
is that valid data are those that enable a researcher to infer underlying
phenomena. The distinction between ‘data’ and ‘phenomena’ is made much
 Data, theory, and applications 675

of by Gregg (1993). Citing Bogen and Woodward (1988), Gregg emphasizes

that theory construction is concerned with phenomena (i.e. some generalized
aspect or event such as the melting point of lead) and not with data (i.e. par¬
ticular observations such as the actual measurements of the temperature at
which samples of lead melt). The question that arises, of course, is how
‘phenomena’ can be derived from ‘data’. On this Gregg has nothing to say.
The answer customarily given is that data provide evidence of phenomena
when the results they provide are consistent. To demonstrate consistency, it is
necessary to show that the data are not just a reflection of the instrument used
to collect them. This can be achieved by comparing the results obtained from
one set of data with those obtained from another—ideally, some ‘baseline’
such as natural language use. We have seen recourse to this notion of concur¬
rent validity in Beebe and Cumming’s evaluation of data obtained by means
of a Discourse Completion Questionnaire and in Birdsong’s (and others’) jus¬
tification of metalingual judgement data. There is, however, a logical objec¬
tion to such a procedure, of which Birdsong (1989) is fully aware:

One must be aware that not all tasks, linguistic or metalinguistic, tap the
same source of linguistic knowledge. Thus the results of one type of test
may not be useful in proving the concurrent validity of another source
(1989: 118).

In effect, Birdsong is acknowledging the position adopted by variabilists and,

in fact, he explicitly refers to the work of one variabilist, Bialystok. Learners
vary in the way they use their linguistic knowledge in accordance with a vari¬
ety of linguistic, situational, and psycholinguistic factors (see Chapter 4). It is
to be expected, therefore, that data collected from one source will not com¬
pletely match those collected from another.
If data cannot be validated concurrently, where does this leave us? One
possibility, advanced by the variabilists, is to argue that competence itself is
variable, but this solution is not acceptable to many. Another solution is to
acknowledge that different types of data will serve different purposes and
shed light on different aspects of L2 acquisition—a view that is compatible
with the analogue of the prism in the introduction to Part Seven of this book.
The data needed to investigate the kinds of specific hypotheses derived from
the Competition Model or UG theory are only readily available through ex¬
perimental elicitation. The data needed to investigate the claims of the Multi¬
dimensional Model need to reflect natural language use. However, while
admitting that ‘good data’ are, above all, relevant data, we also need to recog-
nizeThe dangers of highly controlled data for, as Hatch, Shirai, and Fantuzzi
(1990) have pointed out, they may mislead because the results they provide
fail to generalize to other types of data. At the very least, then, researchers
need to study whether consistency is present and, when it is not, acknowledge
the problem and search for an explanation.
676 Conclusion

It follows that multiple data sources are needed. This is, in fact, what we
find most researchers currently recommending and many actually practising.
Good data, then, are data that are relevant to the particular descriptive or the¬
oretical goals of the research and which are compared with other data. Good
research is research that makes use of multiple sources of data, that gives re¬
cognition to the limitations of the data sources used, and, in Birdsong’s
words, recognizes that ‘each method carries with it impediments to the trans¬
lation of data to theory’ (1989: 613).

The construction and role of theories in second language

acquisition research
It would be a mistake to suggest that all SLA research is or should be directed
at theory development. Gregg (1993) points out that much research in any
field is usefully directed at measuring phenomena, constructing new instru¬
ments, or creating new phenomena. Nevertheless, many researchers are in¬
creasingly committed to providing an explanation of how an L2 is acquired.
While much of the earlier work was primarily concerned with describing the
object of enquiry (the language produced by L2 learners), subsequent re¬
search has been directed at developing theories to explain the wide range of
phenomena that have been discovered and to predict new phenomena. In the
1980s, in particular, theories appeared at a startling rate. It is not surprising,
therefore, that some L2 acquisition researchers have felt the need to ask a
number of basic questions about theory-building in their discipline, drawing
on recent work in the philosophy of human science (for example, Darden
1991 and Laudan 1990), which seeks an understanding of how research pro¬
grams are actually conducted.7 The main questions are as follows:
1 What approach to theory-building should SLA research adopt?
2 What should be the scope of a theory of L2 acquisition?
3 What form should a theory take?
4 How can different theories be evaluated?

Approaches to theory building in SLA research

Long (1985a) draws on the work of Reynolds (1971) to identify two broad
approaches to theory building: the ‘research-then-theory’ approach and the
‘theory-then-research’ approach. He suggests that a research-then-theory
approach involves the following four activities:
1 Select a phenomenon and list all its characteristics.
2 Measure all the characteristics in as many and as varied situations as
possible.
3 Analyse the resulting data by looking for systematic patterns.
4 Formalize significant patterns as theoretical statements (the laws of
nature).
 Data, theory, and applications 677

This provides an accurate description of much of the early SLA research. The
phenomenon under study was ‘learner language’, its characteristics (errors,
developmental features, variability, etc.) were examined, and ‘systematic
patterns’ were identified. This led to the formulation of a number of ‘laws’ in
the form of generalizations about L2 acquisition (for example, ‘in the early
stages of development learners tend to simplify by omitting both grammatical
functors and content words’). It would seem, though, that there is also the
possibility of a fifth step, not mentioned by Reynolds or Long:

5 Gather together the ‘theoretical statements’ into an explicit theory ex¬

pressed in axiomatic or causal process form.

This step is evident in the theories that came out of the descriptive work of the
1970s—for example, ‘creative construction theory’ or the ‘variable compet¬
ence model’. It constitutes an optional rather than a required step. In Gregg’s
(1993) view, it results in ‘shallow theories’ (i.e. theories that stick as closely as
possible to what is observable and, therefore, do not address ‘phenomena’).
The research-then-theory approach is characterized by Reynolds (as in
Long 1985a: 390) in terms of the following activities:

1 Develop an explicit theory in axiomatic or causal process form.

2 Derive a testable prediction from the theory.
3 Conduct research to test the prediction.
4 If the prediction is disconfirmed, modify the theory and test a new pre¬
diction (or abandon the theory altogether).
5 If the research findings confirm the prediction, test a new one.

It is clear, then, that what is the possible end point of a research-then-

theory approach can serve as the starting point in a theory-then-research ap¬
proach. This approach, however, also lends itself to the construction of ‘deep
theories’ (i.e. theories that go beyond what has been observed by acknowl¬
edging the importance of those aspects of nature that cannot be observed).
Importantly, such theories do more than just explain; they also allow for pre¬
dictions to be formulated, tested, and modified. One possibility afforded by
this approach is the adoption of a theory for step (1) from outside the realm of
SLA research. This has increasingly happened: for example, the skill-building
models adopted from cognitive psychology or the linguistic models taken
from theoretical linguistics.
It is unnecessary and, indeed, unwise, to try to argue that one approach to
theory building is superior to the other. Indeed, the two approaches can inter¬
lock in a cyclical endeavour. Long (1985a) rightly recognizes that both ap¬
proaches have their strengths and weaknesses. He argues that researchers in
the research-then-theory approach are less likely to be ‘wrong’ because their
theoretical claims are based firmly on empirical evidence—always providing,
of course, that their observations are valid and reliable—but the end result of
their efforts may be ‘limited’ and perhaps ultimately ‘irrelevant’, and, in
678 Conclusion

Gregg’s eyes, result in a ‘shallow’ theory. However, there would seem to be no

reason why a ‘shallow theory’ should not develop into a ‘deep theory’ over
time, as in fact often happens when theorists in this tradition go beyond the
empirical evidence. Long argues that researchers in the theory-then-research
tradition are able to offer an ‘approximate answer’ right away and also are
more likely to bring about a paradigm shift, as it is this approach that is asso¬
ciated with ‘scientific revolutions’. He recognizes, however, that they tend to
resist abandoning their theories, preferring, if necessary, to modify or add
parts to them in order to immunize them against disconfirming evidence.
These approaches reflect the two ways in which SLA research can proceed.
Beretta (1991: 505) refers to them as ‘bottom-up and top-down strategies’.
He argues that the former requires researchers to seek plentiful data before
they formulate any theory, and intimates that researchers should resist the
temptation to formulate an explicit theory prematurely. According to
Beretta, a top-down strategy requires the researcher to formulate a provi¬
sional theory immediately so that this can then guide enquiry. It is probably
true to say that most L2 researchers accept that both strategies are legitimate
and useful, although some, like Gregg (1989) have sufficient belief in one par¬
ticular theory (Universal Grammar) to argue forcefully for a theory-first ap¬
proach and against a data-led one. Beretta also notes that there are
advantages in the advocacy of a ‘tribal model’ (a theory that it is aggressively
promoted so that it stimulates others to explore and elaborate it). The prob¬
lem with a theory-first strategy, of course, is that it requires a theory of suffi¬
cient strength to command allegiance, and it is this which, in the eyes of many
researchers, is still lacking.

The scope of a theory of second language acquisition

Second language acquisition is an enormously complex phenomenon—so
complex, perhaps, that it is not yet clear exactly what a theory of L2 acquisi¬
tion needs to account for. One way of specifying the scope of such a theory
might be to start by identifying the facts of L2 acquisition. This is the ap¬
proach that Long (1990c) adopts. He lists what he calls the ‘well-established
findings’ of SLA research (1990c: 656), and argues that these constitute ‘the
least that a second language acquisition theory needs to explain’. Such an ap¬
proach makes obvious sense and, incidentally, provides a justification for
such a book as this one, which has as one of its major goals a review of what
has been discovered about L2 acquisition. Nevertheless, it also has problems.
The first is that there is often no agreement about what constitutes the ‘well-
established findings’. While it is often possible to agree on broad generaliza¬
tions (for example, ‘age differences systematically affect how fast learners
learn’), it is less easy to reach agreement on more specific statements (such as
‘the “critical period” for the acquisition of a native-like phonology ends at the
age of six’). Second, there is the question of deciding how many of the well-
 Data, theory, and applications 679

established findings a theory needs to account for. Long, for instance, states
that ‘a theory must account for at least some of the major accepted findings’
(1990c: 660, my emphasis added), inviting questions as to what ‘some’ and
‘major’ mean, questions for which, not surprisingly, no answers are provided.
It is probably true to say that, to date, there is no agreement in SLA research
about what facts or how many facts need to be accounted for. This, above all,
is why the field is characterized by a multiplicity of theories.
Three views regarding the scope of an L2 acquisition theory can be iden¬
tified. One is that the field should strive towards a single, comprehensive
theory, or at least seek to eliminate inferior theories. The second view is that a
modular approach will work best, informed perhaps by an over-reaching
framework that enables researchers to identify fairly specific domains in
which to theorize. The third view is that multiple, overlapping theories are
inevitable and even, perhaps, desirable.
Probably the most substantial attempt to construct a comprehensive theory
is that of Spolsky (1988; 1989). This theory, which Spolsky refers to as a ‘gen¬
eral theory’, takes the form of a set of statements that describe the ‘conditions’
under which L2 learning can take place. As we saw in Chapter 11, these con¬
ditions are of various kinds—’necessary’, ‘typical’, and ‘graded’—depending
on the nature of the relationship between a particular condition and a par¬
ticular outcome. Because many of the conditions are of the ‘typical’ and
‘graded’ kind, Spolsky refers to his theory as a ‘preference model’. The model
contains a total of 74 statements of conditions, which Spolsky claims are ‘the
natural and logical conclusion of current research in second language learn¬
ing’ and ‘a description of the state of the art’ (1988: 384). In addition, Spolsky
(1989: 28) provides a schematic outline of the different components of the
model. This is important because it provides a basis for grouping the various
conditions (for example, under such headings as ‘social context’, ‘attitudes’,
and ‘learning opportunities’) and also because it indicates the nature of the re¬
lationship among the various components.
A comprehensive theory such as Spolsky’s general theory is attractive to
SLA research because it affords a basis for systematic enquiry. It provides a
framework in which individual researchers can locate their specific lines of
enquiry, it allows for specific predictions based on the conditions to be tested,
and it provides a blueprint for investigating the whole of L2 acquisition (i.e.
for determining which factors contribute to the universal and variant proper¬
ties of L2 acquisition and to what extent they do so). In other words, it helps
to shape the whole field and caters for research with both micro and macro
goals. Such a comprehensive theory has a number of drawbacks, however.
First—and perhaps most serious—the interrelationships among the different
sets of statements (i.e. the components of the model) are underspecified, as
to specify them fully would run the risk of producing a model of unwieldy
complexity. Spolsky’s theory is, in the final analysis, little more than a list of
more or less unrelated statements, with all the limitations that this entails.
680 Conclusion

Second, there is the danger—acknowledged by Spolsky—that the statements

will be pitched at such a high level of generality as to be of doubtful value.
One wonders, for example, about the explanatory value of conditions such as
the ‘Attitude Condition’, which blandly states that ‘a learner’s attitudes affect
the development of motivation’ (Spolsky 1989: 23). Finally, it is not clear
how the whole theory can be falsified (although, as we will shortly see, the no¬
tion of ‘falsification’ is itself problematic). It may be possible to falsify the spe¬
cific conditions, but how many conditions have to be shown to be false before
the whole theory becomes untenable? Spolsky’s general theory results in a
‘messy, fuzzy, overlapping picture’ (Hatch, Shirai, and Fantuzzi 1990: 711).
Some researchers find this tolerable because of the importance they attach to
forming a whole picture.
Many researchers, however, prefer a modular approach to theory building.
A good example of such an approach is that found in White (1989a) and dis¬
cussed in Chapter 10 (see Figure 10.4, page 439). The case for modularity can
be made on both theoretical and procedural grounds. In the case of White and
other researchers in the UG tradition, there are logical arguments for claiming
the existence of a ‘language module’ that can be investigated independently of
other domains (such as perceptual modules). UG researchers stake out a
claim for the necessity of examining the language module separately from
other modules. However, other researchers may choose to investigate a spe¬
cific domain simply because it is procedurally convenient to do so. For ex¬
ample, Gardner’s work on attitudes and motivation (see Chapters 6 and 11)
has proceeded without reference to the existence of developmental universal
in learner language, probably because the psychometric tradition, to which
Gardner adheres, has employed data collected by means of tests and ratings,
which do not readily afford a developmental perspective.
For a theory to be truly modular, it is necessary to specify (1) the domain
covered by the theory, as narrowly as possible, and (2) the domains that are
not covered by the theory, in broad outline. Many theories are modular in the
very restricted sense that they tackle only part of the picture, but they neither
state explicitly what domain they seek to explain, nor do they provide any in¬
dication of the domains not covered by the theory. UG-based theories,
though, can legitimately call themselves ‘modular’ as they conform to both
the above requirements.
One of the dangers of modularity is unwarranted reductionism (i.e. the do¬
main of investigation is delimited over-narrowly and without theoretical jus¬
tification). Another serious problem is atomism (i.e. the investigation <af one
domain takes place in total isolation from the study of other domains). An ex¬
planation may satisfactorily account for a delimited set of phenomena and yet
be found wanting when these same phenomena are viewed as part of the com¬
plete picture. As Hatch, Shirai, and Fantuzzi put it, ‘to make our research
feasible, we try to limit our investigations to one area, but in so doing, we may
advance explanations that are faulty’ (1990: 702).8 This led them to argue
 Data, theory, and applications 6 81

that ‘partial theories are misleading’. Ultimately, therefore, a modular theory

has to be tested in relation to the whole picture. It is valid in the extent to
which its explanation of a single domain holds true when this domain is
considered in relation to other domains.
Most theories of L2 acquisition are neither comprehensive nor truly modu¬
lar. Rather they tackle a particular area or adopt a particular perspective (of¬
ten derived from a parent discipline—cognitive psychology, social
psychology, sociolinguistics, linguistics, neurolinguistics, education) without
reference to other areas or perspectives. A multiplicity of theories is the result,
as is amply demonstrated in this book (see Part Four in particular) and in
other survey books. In Ellis 1985a I reviewed seven theories, McLaughlin
(1987) considers five, while Larsen-Freeman and Long (1981) also examine
five (although not the same five as McLaughlin). Furthermore, this theoretical
pluralism shows no signs of abating, for as Spolsky has observed ‘new theo¬
ries do not generally succeed in replacing their predecessors, but continue to
coexist with them uncomfortably’ (1990: 609).
The question that arises is to what extent this multiplicity is to be viewed
positively or negatively. Beretta (1991) notes that opinions differ consider¬
ably on this point. On the one hand there are the ‘relativists’ like Schumann
(1983) who view SLA research as an ‘art’ rather than a ‘science’, and argue
that ‘in art perspectives are neither right or wrong; they are simply more or
less appealing to various audiences’ (1983: 66). On the other hand, there are
those like Griffiths (1990), Long (1993), and, to a certain extent, Beretta him¬
self, who see multiple theories as problematic for SLA research and consider
the elimination of some theories in favour of others a necessary goal if the
field is to advance. Long points out that whereas some theories can be seen as
complementary, others are clearly oppositional. He gives examples of rival
claims in choice of domain (for example, UG-based theories v. variabilist the¬
ories), within a single domain (for example, the conflicting arguments regard¬
ing the availability of UG in adult learners), and with regard to the specific
variables considered important by mentalist, environmentalist and inter-
actionist theories. He argues that where opposition exists, ‘culling’ needs to
take place, pointing out that the history of science shows that successful sci¬
ences are those that are guided by a ‘dominant theory’. Culling calls for the
identification of assessment criteria. Before we turn to this issue, however, we
will consider what form theories of L2 acquisition can take.

The form of a theory of second language acquisition

There has been far less discussion about the form that a theory of L2 acqui¬
sition should take. Crookes (1992) and Gregg (1993) distinguish ‘property the¬
ories’ and ‘transition theories’. The former are concerned with providing
accounts of static systems; they are ‘ways of representing dispositions, com¬
petencies, or bodies of knowledge’ (Crookes, 1992: 433). Such theories are
682 Conclusion

useful in that they help to identify the states of L2 knowledge and the com¬
ponents of such states to be found in learners at different stages of their devel¬
opment. Much of the modelling of L2 knowledge found in UG-based theories
of L2 acquisition is of this nature (see, for example, Figure 8.1 or, again, Fig¬
ure 10.4). But there is general recognition that property theories in themselves
cannot satisfactorily explain L2 acquisition. A transition theory that seeks to
account for how changes in the state of a system take place is also needed.
Crookes argues that an adequate theory must specify the ‘mechanisms’ that
are responsible for such changes. Long (1990c: 654) defines ‘mechanisms’ as
‘devices that specify how cognitive functions operate on input to move a
grammar at Time 1 to its new representation at Time 2’, although this prob¬
ably needs to be broadened to include mechanisms of a more affective nature
such as motivation. A satisfactory theory, then, must identify the mechan¬
isms—cognitive and affective—responsible for acquisition.
How, then, should this be done? Crookes points out that traditionally the¬
ories have taken the form of a series of deductively related sentences, often
utilizing a highly formalist language such as that provided by mathematics or
logic. However, this kind of theory has never figured strongly in SLA re¬
search, which instead has preferred what Crookes calls ‘a statement-picture
complex’ form. According to this, a theory consists of some kind of pictorial
element (the ‘model’) and a supporting set of statements containing general¬
izations based on and supporting the model. The ‘model’ provides an iconic
element and is likely to employ an analogy of some kind, which may or may
not be explicitly acknowledged. For example, Crookes suggests that Krash-
en’s Monitor Theory incorporates a ladder analogy (in the natural order hy¬
pothesis) and a learner-as-sponge analogy (in his account of how ‘acquisition’
takes place). Schumann (1983), with tongue only partly in cheek, discusses a
number of pictorial models of L2 acquisition from the perspective of a ‘cur¬
ator of an exhibition of SLA art’ (1983: 67) to make his point that theories
need to be considered from an aesthetic point of view. The supporting state¬
ments, Crookes suggests, need to be clear and explicit, which can be achieved
through the appropriate use of formalisms, although few L2 acquisition
theories have attempted this.

Evaluating second language acquisition theories

Three approaches to theory evaluation can be identified. One consists of

identifying a set of criteria that can be applied to all theories as a way of elim¬
inating those that are unsatisfactory. This, according to Beretta (1991) and
Long (1993), should be the goal of SLA research. The second approach is to
abandon any attempt to evaluate theories on rational grounds, and to turn in¬
stead to aesthetic criteria. The third approach is to accept that theoretical
pluralism is not just a temporary feature of an immature discipline, but is here
to stay and to try to avoid the attendant problems of absolute relativism by
 Data, theory, and applications 683

evaluating theories in relation to their particular contexts and purposes. We

will briefly consider these three approaches.
A number of criteria are traditionally invoked to evaluate theories.
McLaughlin (1987; 1990c), for instance, mentions the following:

1 Norms of correspondence: the extent to which the theory fits the facts that
it seeks to explain
2 Norms of coherence: the extent to which the theory fits the body of know¬
ledge that has already been established and is consistent with other related
theories
3 Practicality: the extent to which a theory is ‘heuristically rich’ in stimulat¬
ing and guiding research (McLaughlin 1990c: 619)
4 Falsifiability: the extent to which the theory affords hypotheses that can be
disconfirmed.

In particular, McLaughlin emphasizes the importance of falsifiability.

However, this has recently come under attack. Beretta (1991) argues that it is
problematic because of the impossibility of obtaining neutral, objective data
with which to test theoretical claims and, also, because L2 acquisition
theories tend to be formulated in such a way that they allow for ad hoc or
auxiliary hypotheses as a way of immunizing initial hypotheses against dis-
confirming evidence. Schumann (1993) makes the same point when he points
out that it is very difficult to test a hypothesis in isolation because every
hypothesis is embedded in ‘a network of auxiliary assumptions’. Thus, when
the results fail to support the hypothesis, we cannot tell whether this is
because the hypothesis is wrong or because one or more of the auxiliary
hypotheses is wrong. He argues that ‘falsification is ... extremely difficult, if
not impossible to achieve’ (1993: 296).
Long (1993), however, sees greater merit in falsifiability, although he also
acknowledges the problems. He draws on Darden’s (1991) work on how the¬
ory changes in genetics to suggest five sets of ‘assessment strategies’:

1 Criteria applicable before the empirical testing of a theory (for example, in¬
ternal consistency)
2 Criteria for assessing the empirical adequacy of a theory (for example, ex¬
planatory adequacy and generality)
3 Criteria for assessing a theory’s future potential (for example, fruitfulness)
4 Consistency with accepted theories in other fields
5 Metaphysical and methodological constraints (for example, experimental
testability).

Long suggests that these criteria provide a basis for theory evaluation but he
also accepts ‘that there is no consensus on a universal set of criteria at present’
(1993:239).
684 Conclusion

Indeed, Beretta (1991) concludes his excellent discussion of the various

evaluative criteria with the assertion that there are no foolproof, indispen¬
sable criteria available. However, he goes on to argue that SLA research must
find ways of eliminating theories to achieve maturity, and advances the
‘modest’ notion that ‘evidence’ (in the form of data that fail to confirm a
hypothesis or unsuccessful attempts to replicate findings) is capable of
changing our views. In short, although it is not logically possible to falsify
hypotheses, we must (and do) act as if it is. Long (1993) also suggests that we
need to act as if a set of universally valid assessment criteria exists. He argues
that different criteria may be important at different stages of theory
development, and that the failure to identify a universal set of criteria may
reflect a ‘problem in the timing of their application, not conflicts in principle’
(1993: 242). He goes on to suggest that criteria, like theories, should be
subjected to empirical testing and adjusted when they are found to be ‘false’.
It would seem, though, that the same problems that exist with regard to
demonstrating the falsifiability of theories are also likely to hold where
assessment criteria are concerned.
There are other possible approaches to theory evaluation. Schumann has
advanced two. Schumann (1983) makes much of the fact that scientists
working in the physical sciences now acknowledge that there is no ‘objective
reality’. The recognition that ‘we create the reality we study’ leads to an
acceptance of ‘philosophical flexibility’ (1983: 51) and to treating science as
‘art’. Clearly, a very different set of criteria are called for if this perspective is
adopted. Citing Kramer (1980), Schumann suggests that ‘innovation’, ‘tone
of voice’, and ‘metaphor’ can be used to evaluate theories, and goes on to
apply these criteria to a number of L2 acquisition theories. This approach to
theory evaluation is attractive, not least because it may reflect what many of
us covertly do when we consider theories, but ultimately it is unsatisfactory
because it denies any role for rationality in theory development. Ultimately a
theory is not ‘good’ because it is ‘beautiful’, but also because it ‘makes sense’.
In a later article, Schumann (1993) suggests that theory development
should be seen as a process of ‘exploration’. This refers to ‘efforts to expand,
revise, alter, and ultimately to understand and assess the validity of the
construct’ (1993: 301). Schumann suggests that one way in which explora¬
tion can take place is by trying to understand a phenomenon at a different
level of organization or in relation to another field. For example, socio-
affective theories, which are the main focus of his own research (see Chapter
6), can be explored in terms of social/psychological factors and through
neurological enquiry. Schumann sees a ‘reductionist’ approach that accom¬
modates explanation at different levels as preferable to a ‘closurist’ approach
of the kind favoured by Long.
Schumann’s reductionism comes close to relativism (the belief that any
theory is as good as another and that, therefore, culling is neither desirable
nor possible). There is considerable resistance to such a position, however
(see the papers by Beretta and Long referred to above). The problem is this:
how can
 Data, theory, and applications 685

we avoid relativism if there are no agreed criteria for choosing among theor¬
ies? One answer might be to adopt the third approach—to evaluate theories
in relation to the context in which they were developed and the purpose(s)
they are intended to serve. A UG-based theory, for example, is to be un¬
derstood in terms of the field of Chomskyan linguistics from which it was de¬
veloped, and needs to be evaluated with regard to the contributions that it
makes to this field. It would be entirely inappropriate to try to understand and
evaluate such a theory from the point of view of a foreign language teacher. In
contrast, the Variable Competence Model (Ellis 1984a) was designed to ac¬
count for the kind of variable performance which I observed in my longitud¬
inal study of three language learners in a London language unit. It would not
be appropriate to evaluate such a theory in terms of the contributions it can
make to linguistic theory. This perspective also allows for variation in the
form that theories take; for example, whereas some theories will need to pro¬
vide very precise statements couched in formalist language, others may be
better served by statements of a more general nature.
An approach to evaluation that acknowledges that theories are contex¬
tually determined allows for an acceptance of complementarity without a
commitment to absolute relativism, for it can still be argued that among
theories constructed for the same purpose and context, one does a better job
than another because it is more complete, fits the facts better, affords more
interesting predictions, is more consistent with other theories, etc. Bialystok’s
Theory of L2 Learning (see Chapter 9), for example, might be considered a
better theory than Krashen’s Monitor Theory because it allows for an inter¬
face between explicit and implicit knowledge, for which there is empirical
support (for example, Pienemann 1989). But theories designed to meet differ¬
ent purposes in different contexts will be allowed to co-exist.
We noted earlier that theories of L2 acquisition are great survivors. Theor¬
ies in SLA research are-not usually dismissed as a result of empirical study or
powerful argumentation but, instead, tend to slip slowly and gently into obli¬
vion. Thus, over the years, it is possible to detect a gradual waning of influ¬
ence of some theories (such as the Contrastive Analysis Hypothesis or
Monitor Theory), although many of their ideas live on and are incorporated
into other models. This is a reflection, perhaps, of the general recognition that
L2 acquisition is extremely complex, and therefore does not lend itself to a
single explanation. Different explanations are seen as useful because they fo¬
cus attention on different sides of the prism—to invoke again the image with
which we began this section. The multiplicity of theories also reflects the fact
that different theories appeal aesthetically to different people and, more cru¬
cially, that theories are developed in (and for) a particular context, and that
the context shapes the content and form that a particular theory takes.
To sum up, we have seen that there is considerable disagreement among
SLA scholars about theories. This is evident in the role of theory in SLA re¬
search, the scope of a theory, the form it should take, and how it should be
686 Conclusion

evaluated. We can detect two poles, with many shades of opinion in between.
At one pole there is ‘a healthy and unusually polite acceptance of the possibil¬
ity of pluralism in the answers proposed, a willingness to concede that differ¬
ent models might be needed for different aspects of the problem, an
acceptance that different points of view might lead to different theories’
(Spolsky 1990: 613). At the other there is the belief that research should fol¬
low the assumed methods of the hard sciences, with no room allowed for
complementarity or personal preference (Griffiths 1990). At the moment the
pluralists are winning out over the closurists. This is perhaps as it should be;
those theories that are found useful by researchers and practitioners (such as
teachers) for their varying purposes will continue to flourish.

The application of second language acquisition research to other

fields of enquiry
On a number of occasions in the preceding chapters, we have noted that the
study of L2 acquisition may be of value to other disciplines. One of the most
obvious ways in which it can be of service is by providing a rich body of data
that can be used to address theoretical issues of importance in the ‘parent’ dis¬
ciplines. For example, the study of how L2 learners grammaticalize their in¬
terlanguages (see Chapter 9) can inform functional theories of grammar.
Similarly, experimental studies designed to test UG-based hypotheses can be
used to refine the model of grammar on which they are based. SLA research
and linguistics enjoy a symbiotic relationship, for as Gass (1989) has put it:

... facts of second-language learners force us to look deeper into questions

of language. Facts of language force us to consider them in the light of se¬
cond language learners. Whether it is the chicken first or the egg first comes
largely from one’s major interest and not in anything inherent in the
chicken or the egg (1989: 526).

We might note, though, that this symbiosis is not without its problems. It is
not clear, for instance, what should give way—the ‘facts of second-language
learners’ or the ‘facts of language’—when the results of empirical studies of
L2 learners are not in conformity with the predictions of linguistic theory.9
The main area of application, however, is probably L2 pedagogy. As Spol¬
sky notes, ‘we have a traditional concern to consider not just the explanatory
power of a theory but also its relevance to second language pedagogy’ (1990:
610). This is reflected in a series of articles over the years that have addressed
the relationship between SLA research and second/foreign language teaching
(for example, Tarone, Cohen, and Dumas 1976; Hatch 1978e; Corder 1980;
Lightbown 1985; Widdowson 1990; Long 1990b; Nunan 1990b and 1991;
Ellis 1993c). In these articles we can identify a number of different positions:
 Data, theory, and applications 687

1 The results of SLA research cannot be safely applied to language ped¬

agogy because they are too uncertain.
2 SLA research provides a basis for teacher ‘education’ but not for teacher
‘training’. That is, it can help teachers develop reasonable expectations
about what they can achieve in their teaching, but cannot be used to tell
them how to teach.
3 SLA research provides information and actual data that can be used in
the construction of tasks designed to raise teachers’ awareness of the
likely relationship between teaching/learning behaviours and L2
acquisition.
4 The results of SLA research (and, in particular of classroom-oriented re¬
search) provide ‘hard evidence’ which should be used to advise teachers
about what techniques and procedures work best.

As we can see, these position range from a super-cautious ‘don’t apply’ to a

confident ‘go ahead and apply’.
The cautious ‘don’t apply’ position is evident in early articles by Tarone et
al. (1976) and Hatch (1978e). Tarone et al. gave seven reasons why SLA re¬
search fails to provide the teacher with satisfactory guidelines. These in¬
cluded: it was restricted in scope, it had only just begun to investigate the
cognitive processes and learning strategies involved in L2 acquisition, the
contribution of individual variables such as personality and motivation had
not been evaluated, the methodology for both the collection and analysis of
data was still uncertain and few studies had been replicated. Tarone and her
colleagues noted that the practices of teaching and research were very differ¬
ent, for whereas teachers had immediate needs they must meet, researchers
could afford to follow a slow, bit-by-bit approach. Hatch (1978e) is similarly
cautious. She notes that researchers have been over-ready to make applica¬
tions for pedagogy, commenting: ‘... our field must be known for the incred¬
ible leaps in logic we make in applying our research findings to classroom
teaching’.
One answer to the views expressed by Tarone et al. and Hatch would be to
demonstrate that there have been considerable developments in SLA re¬
search, which now it make it possible to apply findings with greater confid¬
ence. It is true that much more is known about some issues (such as the nature
of developmental sequences, language transfer, or learning strategies), but it
is also true that many problems continue to exist, not least because of the data
problems discussed earlier. Not surprisingly, then, doubts have continued
about whether SLA research should serve as a basis for giving advice to teach¬
ers. Lightbown, for example, concludes that ‘second-language acquisition re¬
search does not tell teachers what to teach, and what it says about how to
teach they have already figured out’ (1985: 182). However, Lightbown does
find one role for SLA research. She suggests that a knowledge of the findings
of SLA research will help teachers to have ‘much more realistic expectations
688 Conclusion

about what can be accomplished’. This is a fairly humble claim that might
lead many teachers to ask whether it is really worthwhile going to all the time
and effort of familiarizing teachers with the results of SLA research. The
teacher’s task can be made easier, however, by the provision of short, read¬
able accounts of the main findings. A good example of such an account is
Lightbown and Spada (1993).10
A somewhat stronger version of this second position can be found in Wid-
dowson (1990). Whereas Lightbown feels that SLA research should not be
applied to teaching because of its continuing limitations, Widdowson adopts
a more fundamental position. He argues that the procedures involved in con¬
ducting research and in language teaching are not the same, thus making the
direct application of theory and research to language pedagogy impossible.
For Widdowson, theory and research involve ‘an abstraction’, whereas teach¬
ing takes place in concrete and varied contexts. It follows that ‘application
cannot simply model itself on the procedures of empirical research’ (1990:
60). One conclusion from this line of argument might be that theory and re¬
search are of no use to the teacher. However, this is not the position Widdow¬
son adopts. Instead, he argues that what is needed is an attempt to mediate
between ‘outsider research’ and ‘insider research’ (i.e. actual teaching). This
mediation requires ‘teacher education’ (rather than ‘teacher training’) and
should take the form of discussion of ‘issues of current pedagogic concern’
(1990: 66), informed by ‘generalities’ and ‘principles’ supported by theory
and research. By engaging in a ‘process of pragmatic mediation’, teachers can
examine how their particular pedagogic problems can be addressed. This
view of the relationship between theory/research and teaching, then, emphas¬
izes the inevitability of variable solutions, as teachers seek to utilize the in¬
formation available to them in terms of their own teaching context. It is a far
cry from the fourth position discussed below.
The third position is, perhaps, best seen as an extension of the second. In
Ellis 1993c, I argued that SLA research is a rich source of information and
data which can be used to foster the processes that lie at the centre of teacher
development (i.e. the formation of a language teaching ideology, the acquisi¬
tion of techniques and procedures for action, and the evaluation of these
through reflection). I suggested that one way of making practical use of SLA
research is by developing activities to raise teachers’ awareness about such
issues as the relationship between the questions they ask and L2 learning. The
research, then, provides teachers with ideas at different levels, which they can
then test out in their own classrooms.
Other scholars have shown less reticence in advocating that teachers follow
where the research leads. Corder (1980) sensibly points out that teachers can¬
not wait until researchers have got it right and that it is natural to expect them
to go ahead and make use of the best information available. Long (1990b) is
more forthright still. He argues that teaching and teacher training should be
grounded in ‘hard evidence about what works’. He claims that the teaching
 Data, theory, and applications 689

profession, like the medical profession, needs to be informed by specialized

knowledge unknown to the lay person and to be guided by research informa¬
tion. He is critical of the ungrounded ‘assertions and prescriptions’ found in
teaching manuals. Thus, although, like others, he is aware of the limitations
of the research, he believes that there are sufficient studies affording reliable
results to make it possible to transmit them to teachers in the form of informa¬
tion about what and how to teach.
It is clear that we have come a long way from the uncertainties of the 1970s.
There is now greater confidence in SLA research and more conviction that its
results can inform language pedagogy. However, not all researchers would
agree with Long’s views, not least because they feel that teacher development
does not take place through being told what and how to teach but by explor¬
ing and reflecting on the options available to them. The positions adopted by
Widdowson, Lightbown, and myself, therefore, seem more in accord with
current views of teacher development (see Richards and Nunan 1990). Fur¬
thermore, some educationalists might feel that research undertaken by pro¬
fessional researchers will always be of limited value to language teachers and
that a more worthwhile and exciting approach is action research, where
teachers become researchers by identifying research questions important to
them and seeking answers in their own classrooms (see Nunan 1990b).

Conclusion
In this chapter we have examined a number of issues that are of epistemolog¬
ical and philosophical importance to SLA research. We have restricted our
discussion to those positions reflected in the SLA literature to date. In accord¬
ance with the main aim of this book, no attempt has been made to argue
strongly in favour of particular positions, although my own views have neces¬
sarily intruded. The treatment of the issues has necessarily been somewhat
limited, a reflection perhaps of where SLA research currently stands on them.
It is, however, significant that SLA researchers have felt the need to discuss
these issues in recent years—a sign of a field of enquiry coming of age.
It is probably true to say that SLA research, some twenty-five years after
its inception as an identifiable field of enquiry, is characterized by facts,
opinions, explanations, positions, and perspectives that frequently exist in
an uneasy state of complementarity and opposition. This is not to suggest that
progress has not been made—in broadening the overall scope of SLA
research, in identifying the essential issues in need of investigation, in devel¬
oping methods for studying them, and in collecting an enormous amount
of information about them. Also, the discovery of competing and overlapping
phenomena might be seen as evidence of the principal strength of SLA
research—a willingness to explore a wide range of issues by means of alter¬
native paradigms and methods. No doubt, over time, the pictures provided by
the different sides of the prism will become clearer, but whether this will lead
690 Conclusion

to a single, unifying account of L2 acquisition, as some believe is necessary,

remains to be seen.

Notes
1 Van Lier has pointed out to me that consideration also needs to be given
to applications in the other direction—for example, language pedagogy
can serve as a source of theoretical and empirical knowledge about L2 ac¬
quisition. The applications of such fields as linguistics, sociolinguistics,
cognitive and social psychology, and language pedagogy to SLA research
have been considered throughout this book.
2 Although there has been a substantial amount of work that has examined
the effect of input on comprehension, there have been few attempts to
study how learners’ abilities to comprehend develop, or how they differ
from their abilities to produce (but see Berent 1985). One reason for this
is that comprehension can take place by means of ‘semantic processing’,
involving top-down processing, with the result that it is difficult to obtain
data on learners’ ability to understand specific linguistic features.
3 Natural language use is likely to display considerable variability. It is for
this reason that it invites the kind of contextual analysis undertaken by
variabilists and pragmatists.
4 Larsen-Freeman and Long (1991) argue that there is a difference between
experimental elicitation tasks and tests, reflecting differences in purpose.
Whereas tests measure target-language performance, tasks provide data
about interlanguage language performance. This difference, of course,
may exist for both the researcher/teacher and the learner or for one but
not the other. Arguably, it is the learner’s perspective that is crucial.
5 The telephone calls were not entirely ‘natural’ as they were undertaken by
the researchers with the specific aim of obtaining data about refusals.
However, they did have a communicative purpose and there is no reason
to believe that the persons answering the call treated them as anything but
authentic calls. The distinction between ‘natural language use’ and ‘clin¬
ically elicited language use’ is a fine one.
6 We might also note that the need for a distinction between ‘competence’
and ‘performance’ is not accepted by all SLA researchers (for example, re¬
searchers in the functionalist tradition discussed in Chapter 9).
7 Beretta (1993) points out that whereas earlier work in the philosophy of
human science tended to be prescriptive in nature and, therefore, of lim¬
ited value, more recent work has been directed at achieving an under¬
standing of how actual research is conducted.
8 This is really a restatement of the data problem. One source of data may
afford results that are not supported by other data sources. In particular,
there may be differences between what data elicited experimentally and
natural data show.
 Data, theory, and applications 691

9 This is an obvious problem in UG-based SLA research. If L2 learners fail

to provide evidence of access to a particular principle of UG, this might
suggest (1) that they do not have access to UG or (2) the principle has been
incorrectly formulated in linguistic theory.
10 Lightbown and Spada’s book is about ‘how languages are learned’. In ac¬
cordance with Lightbown’s earlier views concerning the role of SLA re¬
search in education (as opposed to training), this book is directed at
helping teachers to ‘evaluate proposed changes in classroom methodo¬
logy’. It is to be used alongside ‘the teacher’s own experience’ (1993: xiii).

Further reading
D. Johnson, Approaches to Research in Second Language Learning (Long¬
man, 1992) provides a very readable account of different research paradigms
that addresses many of the issues raised in this chapter.
An excellent description of the different data collection instruments can be
found in D. Larsen-Freeman and M. Long, An Introduction to Second Lan¬
guage Acquisition Research (Chapter 2) (Longman, 1991).
Useful articles on theory in SLA research include:
M. Long, ‘Input and second language acquisition theory’ in S. Gass and C.
Madden (eds.), Input in Second Language Acquisition (Newbury House,
1985).
Articles by B. Spolsky, B. McLaughlin, M. Long, and E. Hatch, Y. Shirai, and
C. Fantuzzi in TESOL Quarterly 24: 4.
A. Beretta, ‘Theory construction in SLA: complementarity and opposition’.
Studies in Second Language Acquisition (1991) 13: 493-511.
Articles by M. Long, K. Gregg, and J. Schumann in Applied Linguistics 14: 3.
Three articles indicative of the range of opinions about the application of
SLA research to language pedagogy are:
E. Hatch, ‘Apply with caution.’ Studies in Second Language Acquisition
(1978)2:123-43.
P. Lightbown, ‘Great expectations: second-language acquisition research and
classroom teaching.’ Applied Linguistics (1985) 6: 173-89.
M. Long, ‘Second language classroom research and teacher education’ in C.
Brumfit and R. Mitchell (eds.), Research in the Language Classroom (Mod¬
ern English Publications, 1990).
Glossary

absolute universals

See typological universals.

Accessibility Hierarchy (AH)

The Accessibility Hierarchy is a statement of the markedness of various relat¬

ive pronoun functions (for example, subject, direct object). It lists these func¬
tions in an implicational ordering, such that a given function implies the
existence of all other functions above it. The hierarchy is an example of a typ¬
ological universal and has been widely used as a basis for SLA research.

Acculturation Model

This is a theory of L2 acquisition developed by Schumann (1978a). It treats

L2 acquisition as one aspect of acculturation (the process by which the learn¬
er becomes adapted to a new culture). Various factors influence the social and
psychological ‘distance’ of the learner from the target-language culture and
thereby the rate and ultimate success of L2 acquisition. See also social dis¬
tance and psychological distance.

‘acquisition’

Krashen (1981) uses the term ‘acquisition’ to refer to the spontaneous and in¬
cidental process of rule internalization that results from natural language use,
where the learner’s attention is focused on meaning rather than form. It con¬
trasts with ‘learning’.

Adaptive Control of Thought (ACT) Model

This is Anderson’s (1980; 1983) model of skill learning. The model accounts
for how learners’ ability to perform a skill develops from a declarative stage,
where information is stored as facts, to an autonomous stage, where informa¬
tion is stored as easily accessed procedures. The model views L2 acquisition
as an example of skill-learning.

additional language

The term ‘additional language’ is preferred to ‘second language’ in some set¬

tings (for example, South Africa). This is because it suggests that the L2 will
exist alongside the LI and be of equal but not necessarily greater importance
to the learner.
 Glossary 693

additive bilingualism

See bilingualism.

affective state

The learner’s affective state is influenced by a number of factors, for example,

anxiety, a desire to compete, whether learners feel they are progressing or not.
It is hypothesized that it can influence the rate of L2 acquisition and the ulti¬
mate level of achievement.

analytical strategy

Peters (1977) found that some children seemed to analyse the input into parts.
They manifested systematic development involving first a one-word and then
a two-word stage, etc. Hatch (1974) refers to L2 learners who learn in this
way as ‘rule-formers’. See also gestalt strategy.

anomie

Anomie is experienced by L2 learners (usually in natural settings) who feel

disconnected from the target social group and from their own speech group.
Such learners feel insecure because they believe they do not belong to any so¬
cial group.

anxiety

Anxiety is one of the affective factors that have been found to affect L2 ac¬
quisition. Different types of anxiety have been identified: (1) trait anxiety (a
characteristic of a learner’s personality), (2) state anxiety (apprehension that
is experienced at a particular moment in response to a definite situation), and
(3) situation-specific anxiety (the anxiety aroused by a particular type of situ¬
ation). Anxiety may be both facilitating (i.e. it has a positive effect on L2 ac¬
quisition, or debilitating (i.e. it has a negative effect).

aptitude-treatment interaction (ATI)

See learner-instruction matching.

avoidance

Avoidance is said to take place when specific target-language features are un¬
der-represented in the learner’s production in comparison to native-speaker
production. Learners are likely to avoid structures they find difficult as a res¬
ult of differences between their native language and the target language.
694 Glossary

backsliding

L2 learners are likely to manifest correct target-language forms on some occa¬

sions but deviant forms on other occasions. When this happens they are said
to ‘backslide’. Backsliding involves the use of a rule belonging to an earlier
stage of development. It can occur when learners are under some pressure, as,
for instance, when they have to express difficult subject matter or are feeling
anxious (Selinker 1972).

balanced bilingualism

See bilingualism.

Basic Interpersonal Communication Skills (BICS)

Cummins (1981) uses the term ‘Basic Interpersonal Communication Skills’

(BICS) to refer to the kind of L2 proficiency that learners require in order to
engage effectively in face-to-face interaction. BICS involve the mastery of
context-embedded uses of language in communicative tasks that are relat¬
ively undemanding.

behaviourist learning theory

Behaviourist learning theory is a general theory of learning (i.e. it applies to

all kinds of learning). It views learning as the formation of habits. These are
formed when the learner is confronted with specific stimuli which lead to re¬
sponses, which are, in turn, reinforced by rewards or corrected. Behaviourist
learning theory emphasizes environmental factors as opposed to internal,
mental factors.

bilingualism

Bilingualism refers to the use of two languages by an individual or a speech

community. There are various types of bilingualism. In the case of additive bi¬
lingualism, a speaker adds a second language without any loss of competence
to the first language. This can lead to balanced bilingualism. In the case of
subtractive bilingualism, the addition of a second language leads to gradual
erosion of competence in the first language.

borrowing transfer

See language transfer.

capability

This is the term used by Tarone (1983) to refer to the learner’s actual ability to
use particular rules in language use. Tarone intends it to contrast with com¬
petence. As such, it seems very similar in meaning to L2 proficiency.
 Glossary 695

careful style

This is a term used by Labov (1970) to refer to the language forms evident in
speech that has been consciously attended to and monitored. Such forms are
used more frequently with higher-status interlocutors. A careful style is used
in formal language tasks such as reading pairs of words or doing a grammar
test. See also vernacular style and stylistic continuum.

caretaker talk

When adults (or older children) address young children, they typically modify
their speech. These modifications are both formal (for example, the use of
higher pitch or simple noun phrases) and interactional (for example, the use
of expansions).

casual style

See vernacular style.

change from above

This refers to changes to the dialect spoken by a social group as a result of so¬
cial pressures. Speakers are aware of the change. The linguistic feature in¬
volved in the change can be stigmatized or prestigious. Preston (1989) sees
‘change from above’ as analogous to monitoring in L2 acquisition.

channel capacity

This refers to the language learner’s ability to process utterances in compre¬

hension or production. It involves the ability to recover rules and items from
memory and to use them easily and spontaneously. In the early stages of L2
acquisition, learners have limited channel capacity and slowly develop this at
the same time as they acquire L2 knowledge.

classroom process research

The aim of classroom process research is to subject the teaching-learning be¬

haviours that occur in a classroom to careful observation with a view to
describing as fully as possible what takes place. It views language lessons as
‘socially constructed events’ and seeks to understand how they take place.

closed question

A closed question is one that is framed with only one acceptable answer in
mind (for example, ‘What is the name of the day after Tuesday?’)
696 Glossary

code-switching

Code-switching is one kind of intra-speaker variation. It occurs when a

speaker changes from one variety or language to another variety or language
in accordance with situational or purely personal factors.

Cognitive Academic Language Proficiency (CALP)

Cummins (1981) uses the term ‘Cognitive Academic Language Proficiency’

(CALP) to refer to the kind of L2 proficiency required to engage effectively in
academic study. In particular, CALP involves the ability to communicate
messages that are precise and explicit in tasks that are context-reduced and
cognitively demanding.

cognitive strategies

O’Malley and Chamot define cognitive strategies as learning strategies that

‘operate directly on incoming information, manipulating it in ways that en¬
hance learning’ (1990: 44). They involve such operations as rehearsal, or¬
ganizing information, and inferencing.

cognitive style

Some psychologists consider that individuals have characteristic ways of per¬

ceiving phenomena, conceptualizing, and recalling information. Various di¬
mensions of cognitive style have been identified, including field dependence
and field independence.

communicative competence

Communicative competence consists of the knowledge that users of a lan¬

guage have internalized to enable them to understand and produce messages
in the language. Various models of communicative competence have been
proposed, but most of them recognize that it entails both linguistic compet¬
ence (for example, knowledge of grammatical rules) and pragmatic compet¬
ence (for example, knowledge of what constitutes appropriate linguistic
behaviour in a particular situation).

communication strategy

Communication strategies are employed when learners are faced with the
task of communicating meanings for which they lack the requisite linguistic
knowledge (for example, when they have to refer to some object without
knowing the L2 word). Various typologies of learning strategies have been
proposed.
 Glossary 69 7

comparative method studies

These are studies carried out to measure the different learning outcomes
achieved by two different teaching methods. An example of such a study is the
Pennsylvania Project (Smith 1970), which compared the cognitive-code and
audiolingual methods.

Competence

This term refers to a language user’s underlying knowledge of language,

which is drawn on in actual performance. Theories of language vary in how
they define competence. See also linguistic competence, pragmatic compet¬
ence., and proficiency.

Competition Model

This is a functional model of language use and language acquisition, pro¬

posed initially by Bates and MacWhinney (1982). It views the task of lan¬
guage learning as that of discovering the particular form-function mappings
that characterize the target language. These mappings are viewed as having
varying ‘strengths’ in different languages. For example, in English, case is a
relatively weak signal of agency, whereas in Russian, it is a strong signal.

comprehensible input

‘Input’ refers to language that learners are exposed to. Input that can be un¬
derstood by a learner has been referred to as ‘comprehensible input’ (see
Krashen 1981a). Input can be made comprehensible in various ways: through
simplification, with the help of context, or by negotiating non- and mis¬
understanding. Some researchers (such as Krashen) consider comprehensible
input a necessary condition for L2 acquisition.

comprehensible output hypothesis

‘Output’ is language produced by the learner. It can be comprehensible or

incomprehensible to an interlocutor. Swain (1985) has proposed that when
learners have to make efforts to ensure that their output is comprehensible
(i.e. produce ‘pushed output’), acquisition may be fostered.

concatenative research

In a concatenative approach to the study of individual learner differences, a

general research question serves as a basis for collecting data which then are
used to investigate the relationships among learner factors, and between these
factors and L2 acquisition. Such research is typically correlational in nature.
698 Glossary

consciousness raising

The term ‘consciousness raising’ is used by some researchers with much the
same meaning as ‘formal instruction’ (i.e. an attempt to focus the learner s at¬
tention on the formal properties of the language). In Ellis 1991d, I use this
term with a narrower meaning. I contrast ‘consciousness raising’ with ‘prac¬
tice’, the former term referring to attempts to help learners understand a
grammatical structure and learn it as explicit knowledge. An alternative term
for consciousness- raising is ‘intake enhancement’ (Sharwood Smith 1991).

context

The ‘context’ of an utterance can mean two different things: (1) the situation
in which an utterance is produced—this is the situational context; (2) the lin¬
guistic environment—the surrounding language—this is the linguistic con¬
text. Both types of context influence the choice of language forms, and
therefore have an effect on output. See also psycholinguistic context.

contrastive analysis

Contrastive analysis is a set of procedures for comparing and contrasting the

linguistic systems of two languages in order to identify their structural sim¬
ilarities and differences.

Contrastive Analysis Hypothesis

According to the Contrastive Analysis Hypothesis, L2 errors are the result of

differences between the learner’s LI and the L2. The strong form of the hypo¬
thesis claims that these differences can be used to predict all errors that will
occur. The weak form of the hypothesis claims that these differences can be
used only to identify some out of the total errors that actually occur.

correction

See feedback.

creative construction

Dulay, Burt, and Krashen define ‘creative construction’ as ‘the subconscious

process by which language learners gradually organize the language they
hear, according to the rules they construct to understand and generate sen¬
tences’(1982: 276).

creole

A creole is a pidgin language that has become the native language of a group
of speakers. A creole is generally more linguistically complex than a pidgin
language as a result of a process known as ‘creolization’. This results in a
 Glossary 699

‘creole continuum’, made up of ‘lects’ (i.e. varieties) that vary in complexity

depending on whether they are closest to a pidgin or the standard language.
An example of an English-based creole is Jamaican Creole.

Critical Period Hypothesis

This states that there is a period (i.e. up to a certain age) during which learners
can acquire an L2 easily and achieve native-speaker competence, but that
after this period L2 acquisition becomes more difficult and is rarely entirely
successful. Researchers differ over when this critical period comes to an end.

cross-linguistic influence

This is a term proposed by Sharwood Smith and Kellerman (1986: 1) to refer

to ‘such phenomena as “transfer”, “interference”, “avoidance”, “borrow¬
ing” and L2-related aspects of language loss’. Whereas the term ‘transfer’ is
closely associated with behaviourist learning theory, ‘crosslinguistic influ¬
ence’ is theory-neutral.

debilitating anxiety

See anxiety.

declarative L2 knowledge

Declarative knowledge is characterized by Anderson (1983) as ‘knowledge

that’. In the case of language, it consists of factual information about the L2
that has not yet been integrated or automatized.

descriptive adequacy

An account of a language (i.e. a ‘grammar’) is considered to have ‘descriptive

adequacy’ if it constitutes a complete and explicit description of the rules and
items that comprise native-speaker competence. See also explanatory
adequacy.

developmental feature

See Multidimensional Model,

developmental pattern

See pattern of development

developmental sequence

One of the main findings of L2 research is that learners pass through a series
of identifiable stages in acquiring specific grammatical structures such as neg¬
atives, interrogatives, and relative clauses. To a large extent, although not en¬
tirely, these sequences are not affected by the learner’s LI. See also sequence
of development.
700 Glossary

discourse completion questionnaire

A discourse completion questionnaire provides learners with a description of

a situation designed to elicit a specific illocutionary act (for example, an apo¬
logy) and then asks learners to write down what they would say in such a situ¬
ation or asks them to select what they would say from choices provided. It is
used to investigate learners’ L2 pragmatic competence.

discourse management

When native speakers (or other L2 learners) are addressing L2 learners, they
may seek to modify their speech interactionally in order to avoid communica¬
tion problems. For example, they may restrict the kind of information they
try to convey or use comprehension checks.

discourse repair

When native speakers (or other L2 learners) experience a communication

problem with an L2 learner, they may seek to repair the breakdown (for ex¬
ample, by negotiation of meaning) or to repair a learner error through
correction.

display question

A display question is one designed to test whether the addressee has know¬
ledge of a particular fact or can use a particular linguistic item correctly (for
example, ‘What’s the opposite of “up” in English?’). See also closed question.

educational settings

Whereas many researchers are happy to talk of ‘foreign’ (as opposed to ‘se¬
cond’) language acquisition, others, including the author of this book, prefer
to distinguish different types of language acquisition in terms of context or
setting. A key distinction is between acquisition that takes place in ‘educa¬
tional settings’ (such as schools) and that which takes place in ‘natural set¬
tings’ (such as the street or the work-place).

elaborative simplification

This is a term used by the ZISA researchers (for example, Meisel 1980) to re¬
fer to the simplification that occurs when learners are trying to complexify
their interlanguage systems (for example, through the use of overgeneraliza¬
tion). It contrasts with restrictive simplification.

error

According to Corder (1967), an ‘error’ is a deviation in learner language

which results from lack of knowledge of the correct rule. It contrasts with a
 Glossary 701

mistake. An error can be overt (the deviation is apparent in the surface form
of the utterance) or covert (the deviation is only evident when the learner’s
meaning intention is taken into account). Various frameworks for describing
errors have been developed, including descriptive taxonomies, which focus
on the observable surface features of errors, and surface strategy taxonomies,
which reflect the way in which target language surface structure is altered by
learners.

Error Analysis

Error Analysis involves a set of procedures for identifying, describing, and ex¬
plaining errors in learner language (see Corder 1974). Error Analysis for ped¬
agogical purposes has a long history but its use as a tool for investigating how
learners learn a language is more recent (it began in the 1960s).

error evaluation

Error evaluation involves a set of criteria and procedures for evaluating the
effect that different errors in learner language have on addressees, both native
speakers and non-native speakers. Error evaluation results in an assessment
of error gravity.

error gravity

Error gravity concerns the seriousness of an error. This can be determined

with reference to such criteria as intelligibility, acceptability, and irritation
(Khalil 1985).

error treatment

Error treatment concerns the way in which teachers (and other learners) re¬
spond to learners’ errors. Error treatment is discussed in terms of whether
errors should be corrected, when, how, and by whom.

ethnography

Ethnography makes use of procedures such as detailed observation, inter¬

views, and questionnaires to collect data that are ‘rich’ and that afford mul¬
tiple perspectives. It has been extensively used in the study of bilingual
classrooms.

explanatory adequacy

An account of a language (i.e. a ‘grammar’) is considered to have ‘explanat¬

ory adequacy’ if it not only provides a complete and explicit description of
native-speaker competence, but also provides an explanation for how this
competence is achieved. That is, it must explain how children learn their
702 Glossary

mother tongue. The theory of Universal Grammar seeks to achieve explanat¬

ory adequacy.

explicit L2 knowledge

Explicit L2 knowledge is that knowledge of rules and items that exist in an

analysed form so that learners are able to report what they know. Explicit L2
knowledge is closely linked to metalingual knowledge. It contrasts with im¬
plicit knowledge.

externalized (E) approach

Chomsky (1986a) distinguishes an ‘E-approach’ and an T-approach’ to lan¬

guage study. The former involves the collection of samples of a particular lan¬
guage or languages which are then used to develop-a Tfesertgtive grammar’.
See also internalized (I) approach.

extroversion/introversion

These terms describe the dimension of personality which has been most thor¬
oughly investigated in SLA research. They reflect a continuum: at one end are
learners who are sociable and risk-takers, while at the other end are learners
who are quiet and avoid excitement.

facilitating anxiety

See anxiety.

feedback

In language acquisition, the term ‘feedback’ refers to information given to

learners which they can use to revise their interlanguage. A distinction is often
made between ‘positive’ and ‘negative’ feedback (sometimes referred to as
‘negative evidence’); positive feedback refers to information that indicates a
hypothesis is incorrect. A distinction is also made between ‘on-record’ and
‘off-record’ feedback (Day et al. 1984) depending on whether the feedback is
direct (i.e. a correction) or indirect (for example, in the form of request for
confirmation). On-record feedback supplies ‘direct negative evidence’, while
off-record feedback supplies ‘indirect negative feedback’.

field dependence/independence

Field dependence/independence constitutes one kind of cognitive style. Field-

dependent learners are believed to operate holistically (i.e. they see the field as
a whole), whereas field-independent learners operate analytically (i.e. they
perceive the field in terms of its component parts). Although a number of L2
researchers have made use of this distinction to account for differences in
learners, others dispute its usefulness.
 Glossary 703

foreigner talk

When native speakers address learners, they adjust their normal speech in or¬
der to facilitate understanding. These adjustments, which involve both lan¬
guage form and language function, constitute ‘foreigner talk’. Foreigner talk
has been hypothesized to aid L2 acquisition in a number of ways (for ex¬
ample, by making certain features more salient to the learner).

foreign language acquisition

A number of researchers distinguish ‘foreign language acquisition’ (for ex¬

ample, the learning of French in schools in the United States) and ‘second lan¬
guage acquisition’ (for example, the learning of English by speakers of other
languages in the United States). Other researchers find this distinction prob¬
lematic. See educational settings.

formal instruction

Formal instruction involves some attempt to focus learners’ attention on spe¬

cific properties of the L2 so that they will learn them. Different types of formal
instruction can be distinguished, including inductive (where learners are pro¬
vided with structured input designed to help them learn a rule or item) and de¬
ductive (where learners are given explicit information about a rule or item).

form-function analysis

In a form-function analysis, all instances of a specific linguistic form are

identified in the data and the different meanings realized by this form are
identified.

formulas/formulaic speech

Many learners, particularly in the early stages of L2 acquisition, acquire

ready-made phrases like ‘I don’t know’ and ‘Can I have a_?’. These are
known as formulas. See also routines and patterns.

fossilization

Selinker (1972) noted that most L2 learners fail to reach target-language

competence. That is, they stop learning while their internalized rule system
contains rules different from those of the target system. This is referred to as
‘fossilization’. It reflects the operation of various internal processes.

fragile features

Fragile features of language are those that are acquired late, often with effort,
and only when there is access to adequate input. Examples of fragile features
are plural and tense markings. Fragile features contrast with resilient
features.
704 Glossary

free variation

When a speaker uses two or more variants of a variable structure randomly

(i.e. selects variants without reference to the linguistic or situational context),
free variation occurs. In Ellis (1985c), I argued that free variation arises in
learner language when learners acquire a new form side by side with a previ¬
ously acquired form and use it to realize the same meaning.

frequency analysis

This is the method of analysing learner language that involves identifying the
variants of a given structure and examining the frequency of occurrence of
each variant. For example, a learner may make negative utterances using (1)
‘no’ + verb, (2) ‘don’t’ + verb and (3) auxiliary + verb. A frequency analysis of
the negative utterances produced by this learner would involve counting each
occurrence of the three variants.

frequency hypothesis

The frequency hypothesis states that the order of development in L2 acquisi¬

tion is determined by the frequency with which different linguistic items oc¬
cur in the input.

functionalist model

Functionalist models of F2 acquisition view the task of learning a language as

involving the construction of form-function networks. That is, learners have
to discover which forms perform which meanings in the F2. These models
recognize that learners are likely to construct idiosyncratic form-function
networks (i.e. use forms to perform functions not performed by these forms
in the target language). Functionalist models emphasize the role that com¬
munication plays in the acquisition process.

gestalt strategy

Peters (1977) found that some children in FI acquisition remain silent for a
long time and then suddenly begin producing full sentences. Hatch (1974) re¬
fers to this kind of F2 learner as a ‘data-gatherer’. See also analytical strategy
and silent period.

global error

Global errors are errors that affect overall sentence organization (for ex¬
ample, wrong word order). They are likely to have a marked effect on
comprehension. See also local error.
 Glossary 705

good language learner

Researchers have investigated the individual learner factors that contribute to

L2 learning by investigating what expert, successful L2 learners do in order to
learn an L2. These studies are known as the ‘good language learner studies’.
One of the best known is Naiman et al. 1978.

gradual diffusion model

Gatbonton’s (1978) gradual diffusion model identifies two broad phases in

L2 acquisition: an ‘acquisition phase’, characterized by free variation, and a
‘development phase’ where free variation gives way to systematic variation
and categorical language use. See also variability hypothesis.

grammatical competence

See linguistic competence.

grammaticality judgements

One way of obtaining data on what learners know about the L2 is by asking
them to judge whether sentences are grammatically correct or incorrect. This
method is favoured by some researchers because they believe it provides in¬
formation about learners’ institutions and thus caters for an internalized (I)
approach.

grammaticalization

Some functionalist models of L2 acquisition consider that early L2 acquisi¬

tion is characterized by the use of pragmatic strategies for conveying mean¬
ings that are conveyed grammatically by native speakers. Gradually, learners
move from this pragmatic mode to a grammatical mode as they learn the
grammatical properties of the L2.

hierarchical research

In the hierarchical approach to the study of individual learner differences

(IDs), predictions based on a theory of IDs are made and then tested empiric¬
ally. In fact, this approach has been little used as there is no generally accepted
theory of IDs.

horizontal variation

This refers to the variation evident in learner language at a particular moment

or stage in a learner’s development. It contrasts with vertical variation.
706 Glossary

hypercorrection

Some native speakers over-use a linguistic form which is associated with so¬
cial prestige. For example, some speakers of British English may overextend a
feature like /h/ to words such as ‘hour’, even though this word does not have
this sound in the standard dialect. Preston (1989) has suggested that this pro¬
cess is analogous to over generalization in L2 acquisition.

hypocorrection

This consists of the retention of an old norm in the speech of the working class
because it has covert prestige. Preston (1989) suggests this is analogous to
negative transfer in L2 acquisition.

hypothesis-testing

According to one view of interlanguage development, learners form hypoth¬

eses about the structures of the target language and then ‘test’ these out on the
evidence available from input, with the result that they either accept them or
abandon them and form new hypotheses. This process is not necessarily a
conscious one.

Ignorance Hypothesis

Newmark and Reibel (1968) argue that what is commonly referred to as ‘in¬
terference’ is in fact merely an attempt by the learner to fill gaps of knowledge
by using previous knowledge.

illocutionary act

This term is taken from speech act theory (see Austin 1962). An utterance is
seen as having not only propositional meaning (i.e. saying something about
the world) but also functional meaning (i.e. doing something). An illocution¬
ary act is the functional meaning performed by an utterance. Examples are
‘requesting’ and ‘apologizing’. Searle (1969) has identified various ‘condi¬
tions’ that have to be met in order for a specific illocutionary act to be per¬
formed successfully. For example, for the act of ‘giving an order’ to be
successfully performed, both speaker and hearer must recognize that the
speaker is in a position of authority over the hearer.

immersion education programme

In immersion education programmes the L2 is not taught as a separate sub¬

ject. Rather, it is taught by using it as the medium of instruction for teaching
the content of other school subjects. Immersion education has been widely
used in Canada.
 Glossary 707

implicational scaling

Implicational scaling is a technique used by sociolinguists and some SLA re¬

searchers to represent variation in L2 performance. It rests on the notion that
the presence of one linguistic form in learner language occurs only if one or
more forms are also present. Thus, one form ‘implicates’ other forms.

implicit L2 knowledge

Implicit knowledge of a language is knowledge that is intuitive and tacit. It

cannot be directly reported. The knowledge that most speakers have of their
LI is implicit. The study of linguistic competence is the study of a speaker-
hearer’s implicit knowledge. See also explicit L2 knowledge.

impression management

Impression management concerns the way speakers make use of their lin¬
guistic resources in interaction to create social meanings favourable to them¬
selves. For example, L2 learners may make deliberate use of primitive
interlanguage forms to mitigate the force of threatening speech acts.

incorporation strategy

One way in which learners can construct an utterance is by borrowing a

chunk from the previous utterance and extending it by adding an element at
the beginning or end. See also vertical constructions.

indicator

An indicator is a variable form that is responsive to social factors such as so¬

cial class, age, and gender. See also marker.

indirect negative evidence

See feedback.

individual learner differences (IDs)

The term ‘individual learner differences’ refers to the differences in how

learners learn an L2, how fast they learn, and how successful they are. These
differences include both general factors such as language learning aptitude
and motivation, and specific learner strategies. The differences can be cog¬
nitive, affective, or social in nature.

induced error

Induced errors arise in learner language when learners are led to make errors
that otherwise they would not make by the nature of the formal instruction
they receive.
708 Glossary

inner-directed learners

According to Saville-Troike (1988: 568), inner-directed learners ‘approach

language as an intrapersonal task, with a predominant focus on the language
code’.

Input Hypothesis

The Input Hypothesis was advanced by Krashen (1982) to explain how ‘ac¬
quisition’ takes place. It states that ‘we acquire ... only when we understand
language that contains structure that is ‘a little beyond where we are now’
(1982: 21). Elsewhere Krashen has referred to the idea of input that is ‘a little
bit beyond’ as ‘i + 1’.

instructed language acquisition

This term refers to language acquisition that takes place as a result of at¬
tempts to teach the L2—either directly through formal instruction or indir¬
ectly by setting up the conditions that promote natural acquisition in the
classroom.

instrumental motivation

See motivation.

intake

Intake is that portion of the input that learners notice and therefore take into
temporary memory. Intake may subsequently be accommodated in the learn¬
er’s interlanguage system (i.e. become part of long-term memory). However,
not all intake is so accommodated.

integrative motivation

See motivation.

interactional act

An interactional act is an utterance considered in terms of its structural func¬

tion in discourse, for example, whether it opens, closes, or continues a
conversation.

interactional modification

Interactional modifications occur when some kind of communication prob¬

lem arises and the participants engage in interactional work to overcome it.
They take the form of discourse functions such as comprehension checks, re¬
quests for clarification, and requests for confirmation. See also negotiation of
meaning.
 Glossary 709

interaction analysis

Interaction analysis is a research procedure used to carry out classroom ob¬

servation. It involves the use of a system of categories to record and analyse
the different ways in which teachers and students use language. Various types
of system exist, for example: a category system for coding specific events
every time they occur, a sign system for coding the events that occur within a
predetermined period, and a rating system for estimating the frequency of
specific events.

interactionist learning theory

Interactionist learning theory emphasizes the joint contributions of the lin¬

guistic environment and the learner’s internal mechanisms in explaining lan¬
guage acquisition. Learning results from an interaction between the learner’s
mental abilities and the linguistic input. The term ‘interactionist’ can also be
applied to theories that explain L2 acquisition in terms of social inter¬
action—how communication between the learner and other speakers leads to
L2 acquisition.

interdependency principle

Cummins (1981) formulated the ‘interdependency principle’ to refer to the

idea that cognitive academic language proficiency (CALP) is common across
languages, and can therefore easily be transferred from LI use to L2 use by
the learner. It explains why people who are literate in their LI find fewer
problems in developing CALP in an L2 than those who are not.

interface position

Theories of L2 acquisition that emphasize the distinctiveness of implicit and

explicit knowledge can either maintain that these are completely separate, or
that each knowledge type ‘leaks’, so that explicit knowledge can become im¬
plicit and vice-versa. This latter position is known as the ‘interface position’.
See also skill-building hypothesis.

interference

According to behaviourist learning theory, old habits get in the way of learn¬
ing new habits. Thus, in L2 acquisition the patterns of the learner’s mother
tongue that are different from those of the L2 get in the way of the learning the
L2. This is referred to as ‘interference’. See also language transfer.

Inter-group Theory

This is a theory of L2 acquisition proposed by Giles and Byrne (1982). It char¬

acterizes L2 acquisition as ‘long-term convergence’ and explains it in terms of
710 Glossary

attitudes relating to relationship between the learner and the target-language

community. See Speech Accommodation Theory.

interlanguage

Selinker (1972) coined the term ‘interlanguage’ to refer to the systematic

knowledge of an L2 which is independent of both these learner’s LI and the
target language. The term has come to be used with different but related
meanings: (1) to refer to the series of interlocking systems which characterize
acquisition, (2) to refer to the system that is observed at a single stage of devel¬
opment (‘an interlanguage’), and (3) to refer to particular L1/L2 combina¬
tions (for example, LI French/L2 English v. LI Japanese/L2 English). Other
terms that refer to the same basic idea are ‘approximative system’ (Nemser
1971) and ‘transitional competence’ (Corder 1967).

interlanguage talk

L2 learners often obtain input from other L2 learners. For many learners
interlanguage talk may the primary source of input. In classroom contexts,
interlanguage talk has been referred to as ‘tutor talk’ (Flanigan 1991).

internalized (I) approach

Chomsky distinguishes an externalized (E) approach and an internalized (I)

approach to the study of language. The latter makes use of native-speaker in¬
tuitions about what is grammatical and ungrammatical in order to investigate
the abstract principles that underlie particular grammars. Chomsky favours
the I-approach. See also externalized (E) approach.

inter-speaker variation

In some communities, no speaker has access to all the varieties used by the
community. Instead, one speaker may have access to one variety, another
speaker to a second variety and so on. Inter-speaker variation contrasts with
intra-speaker variation.

intralingual error

Intralingual errors are errors in learner language that reflect learners’ trans¬
itional competence and which are the result of such learning processes as
over generalization. An example might be ‘"'They explained her what to do’.

intra-speaker variation

This refers to the variation in the use of specific linguistic features evident in
the speech and writing produced by individual learners on any one occasion.
This variation reflects linguistic, social, and psychological factors. See also
inter-speaker variation.
 Glossary 711

L1-L2 hypothesis for identity hypothesis)

According to the L1=L2 hypothesis, L2 acquisition is either identical or very

similar to LI acquisition. The similarity may be evident at the level of product
(i.e. in the kind of language produced by the two kinds of learner) or process
(i.e. the mechanisms responsible for acquisition). Views differ considerably
on the validity of this hypothesis.

language acquisition device(LAD)

Mentalist theories of language acquisition emphasize the importance of the

innate capacity of the language learner at the expense of environmental
factors. Each learner is credited with a ‘language acquisition device’ (LAD),
which directs the process of acquisition. This device contains information
about the possible form that the grammar of any language can take. See
Universal Grammar.

language learning aptitude

It has been hypothesized that people possess a special ability for learning an
L2. This ability, known as ‘language learning aptitude’, is considered to be
separate from the general ability to master academic skills, often referred to
as ‘intelligence’. Language learning aptitude is one of the general factors that
characterize individual learner differences. Various tests have been designed
to measure language learning aptitude, for example, the Modern Language
Aptitude Test.

language transfer

Odlin (1989) gives this ‘working definition’ of ‘transfer’:

Transfer is the influence resulting from similarities and differences between
the target language and any other language that has been previously (and
perhaps imperfectly) acquired (1989: 27).
‘LI transfer’ occurs when the ‘influence’ results from the learner’s mother
tongue. Two types of transfer are commonly identified: borrowing transfer
(where the L2 influences the LI) and substratum transfer (where the LI influ¬
ences the L2).

learner-instruction matching

Learner-instruction matching involves an attempt to ensure that the teaching

style is suited to the learner. It is based on the assumption that learners have
different learning styles and that they will learn most effectively if the instruc¬
tion matches their particular learning style. Educational research based on
learner-instruction matching is sometimes referred to as aptitude-treatment
interaction.
712 Glossary

learner strategy

Learner strategies are the behaviours or actions that learners engage in, in
order to learn or use the L2. They are generally considered to be conscious—
or, at least, potentially conscious—and, therefore, open to inspection. See
also cognitive and metacognitive strategies.

‘learning’

Krashen (1981) uses the term ‘learning’ to refer to the development of con¬
scious knowledge of an L2 through formal study. It means that same as expli¬
cit knowledge.

learning strategy

A learning strategy is a device or procedure used by learners to develop their

interlanguages. It is one type of learner strategy. Learning strategies account
for how learners acquire and automatize L2 knowledge. They are also used to
refer to how they develop specific skills. It is possible, therefore, to talk of
both ‘language learning strategies’ and ‘skill-learning strategies’. Learning
strategies contrast with communication and production strategies, both of
which account for how learners use rather than acquire L2 competence. See
also cognitive, metacognitive, and social/affective strategies.

learning style

See cognitive style.

linguistic competence

Researchers who work within the theoretical framework developed by

Chomsky consider it necessary to distinguish competence and performance.
‘Competence’ refers to what speaker-hearers know, and ‘performance’ to the
use of this knowledge in communication. Linguistic competence refers to the
knowledge of the items and rules that comprise the formal systems of a lan¬
guage. It can also be distinguished from pragmatic competence. See also prag¬
matic competence and communicative competence.

linguistic context

See context.

linguistic universals

See typological universals and Universal Grammar,

local errors

Local errors are errors that affect single elements in a sentence (for example,
 Glossary 713

errors in the use of inflections or grammatical functors). They contrast with

global errors.

logical problem of language acquisition

This is a term used by researchers in the Chomskyan tradition. It refers to the

gap between what can logically be learnt from the available input and what
actually is learnt. Input is considered to ‘underdetermine’ language acquisi¬
tion. It is therefore necessary to posit the existence of innate knowledge
of language to account for children’s ability to achieve full linguistic
competence.

Machiavellian motivation

See motivation.

markedness

Various definitions of linguistic markedness exist. The term refers to the idea
that some linguistic structures are ‘special’ or ‘less natural’ or ‘less basic’ than
others. For example, the use of ‘break’ in ‘she broke my heart’ can be con¬
sidered marked in relation to the use of ‘break’ in ‘she broke a cup’. STA re¬
searchers are interested in markedness because it can help to account for
patterns of attested L2 acquisition.

Markedness Differential Hypothesis

This is a hypothesis advanced by Eckman (1977). It makes use of ‘marked¬

ness’ to explain why some LI forms are transferred while others are not. It
claims that learners transfer target-language features that are less marked
than equivalent features in their LI but do not transfer those that are more
marked.

marker

Some variable forms are both ‘social’ and ‘stylistic’. That is, as well as mani¬
festing variation according to social factors such as social class, they also
manifest variation according to situational factors (for example, the
addressee). See also indicators.

mentalist theories of language learning

Mentalist theories of language learning emphasize the learner’s innate mental

capacities for acquiring a language. Researchers in the Chomskyan tradition
consider children (and possibly adults) to possess knowledge of abstract prin¬
ciples of language which serve as a basis for acquiring the rules of particular
languages. Input is seen as a ‘trigger’ that activates these principles. See also
Universal Grammar and LAD.
714 Glossary

metacognitive strategy

Many L2 learners are able to think consciously about how they learn and how
successfully they are learning. Metacognitive strategies involve planning
learning, monitoring the process of learning, and evaluating how successful a
particular strategy is.

metalingual knowledge

Metalingual knowledge is knowledge of the technical terminology needed to

describe language. Metalingual knowledge enables L2 knowledge to become
fully explicit.

mistake

According to Corder (1967), a ‘mistake’ is a deviation in learner language

that occurs when learners fail to perform their competence. It is a lapse that
reflects processing problems. A mistake contrasts with an error.

modality reduction

When L2 learners are under communicative pressure, they may omit gram¬
matical features such as modal verbs and adverbials that are associated with
the expression of modal meanings like possibility and tentativeness. This is
known as ‘modality reduction’.

monitoring

Both native speakers and learners typically try to correct any ‘mistakes’ they
make. This is referred to as ‘monitoring’. The learner can monitor vocabu¬
lary, grammar, phonology, or discourse. Krashen (1981) uses the term ‘Mon¬
itoring’ (with a capital ‘M’) to refer to the way learners use ‘learned’
knowledge to edit utterances generated by means of ‘acquired’ knowledge.

monolingualism

This refers to speakers or speech communities who know and use only one
language—their LI. It can also be characterized as a failure to learn an L2 and
may be associated with a strong ethnic identity and negative attitudes
towards the target-language centre.

morpheme studies

In the 1970s a number of researchers (for example, Dulay and Burt) con¬
ducted studies of a group of English morphemes (for example, V + ing, -ed,
3rd person -s) with a view to determining their order of acquisition. These
 Glossary 715

studies, which were both cross-sectional and longitudinal, were known as

‘morpheme studies’.

motivation

In general terms, motivation refers to the effort which learners put into learn¬
ing an L2 as a result of their need or desire to learn it. In one theory of motiva¬
tion, Gardner and Lambert (1972) distinguish ‘instrumental motivation’,
which occurs when a learner has a functional goal (such as to get a job or pass
an examination), and ‘integrative motivation’, which occurs when a learner
wishes to identify with the culture of the L2 group. Other types of motivation
have also been identified, including (1) ‘task motivation’ or ‘intrinsic motiva¬
tion’—the interest that learners experience in performing different learning
tasks, (2) ‘Machiavellian motivation’—the desire to learn a language that
stems from a wish to manipulate and overcome the people of the target
language, and (3) resultative motivation—the motivation that results from
success in learning the L2.

Multidimensional Model
Meisel, Clahsen, and Pienemann (1981) proposed a model of L2 acquisition
in which a distinction is drawn between developmental and variational fea¬
tures. The former are acquired sequentially as certain processing strategies
are mastered. The latter are acquired at any time (or not at all), depending on
the learner’s social and affective attitudes. The model has been elaborated by
Johnston and Pienemann (1986).

multilingualism
This is the use of three or more languages by an individual or within a speech
community. Frequently, multilingual people do not have equal control over
all the languages they know and also use the languages for different purposes.

natural settings
A natural setting for L2 acquisition is one where the L2 is used normally for
everyday communicative purposes (for example, in the street or the work¬
place). See also educational settings.

naturalistic language acquisition

This refers to language acquisition which takes place in natural settings. It
contrasts with instructed language acquisition.

negative evidence
See feedback.
716 Glossary

negative transfer

See interference.

negotiation of meaning

Communication involving L2 learners often leads to problems in understand¬

ing and breakdown. Frequently, one or more of the participants—the learner
or the interlocutor—attempts to remedy this by engaging in interactional
work to secure mutual understanding. This work is often called ‘negotiation
of meaning’. It is characterized by interactional modifications such as com¬
prehension checks and requests for clarification. Interactionally modified dis¬
course also occurs when there is ‘negotiation of content’, defined by Rulon
and McCreary (1986) as the process by which previously encountered cont¬
ent is clarified.

non-interface position

Theories of L2 acquisition which emphasize the distinctiveness of explicit and

implicit knowledge and which claim that one type of knowledge cannot be
converted directly into the other type adopt a ‘non-interface position’. See
also interface position.

Noun Phrase Accessibility Hierarchy

See Accessibility Hierarchy.

obligatory occasion analysis

This involves identifying contexts that require the obligatory use of a specific
grammatical feature in samples of learner language and calculating the accur¬
acy with which the feature is actually supplied in these contexts (see Brown
1973). See also target-like use analysis.

Observer’s Paradox

According to Labov (1970), good data require systematic observation but

the act of trying to observe contaminates the data collected. Fie refers to this
as the Observer’s Paradox.

open question

An open question is one that has been framed with no particular answer in
mind—a number of different answers are possible. Some questions have the
appearance of being open, but are in fact closed; Barnes (1969) calls these
‘pseudo-questions’.
 Glossary 717

Operating Principles

Slobin (1973) coined the term ‘operating principles’ to describe the various
learning strategies employed by children during LI acquisition. Examples in¬
clude ‘Pay attention to the ends of words’ and ‘Avoid exceptions’. Andersen
(1984a; 1990) has shown how similar principles can be seen at work in L2
acquisition.

order of development

A number of studies of L2 acquisition (for example, the morpheme studies)

have produced evidence to suggest that learners achieve mastery of grammat¬
ical features in a particular order irrespective of their LI or their age. For ex¬
ample, Verb + -ing has been found to be mastered before Verb + -ed.

other-directed learner

According to Saville-Troike (1988: 568), other-directed learners ‘approach

language as an interpersonal, social task, with a predominant focus on the
message they wish to convey’. See also inner-directed learners.

output hypothesis

See comprehensible output hypothesis,

over generalization

Language learners in both first and second language acquisition produce

errors like ‘corned’. These can be explained as extensions of some general rule
to items not covered by this rule in the target language. Overgeneralization is
one type of over-use.

over-use

Over-use involves the use of an L2 feature more frequently than the same fea¬
ture is used by native speakers. It constitutes an ‘over-indulgence’ (Levenston
1971), which may be brought about by differences between the native and
target languages. It may be reflected in errors [overgeneralization) or just a
preference for one target-language form to the exclusion of other possible
target forms.

Parallel Distributed Processing

This is theory of language use and acquisition developed by Rumelhart and

McClelland. Whereas most theories view language as consisting of a set of
rules and items, Parallel Distributed Processing models view it as complex
network of interconnections between ‘units’ that do not correspond to any
718 Glossary

linguistic construct. Learners gradually learn networks with interconnections

of the same ‘strengths’ as those in the networks of users of the target language.

parameter

This is a term used in Chomsky’s theory of Universal Grammar. Some univer¬

sal principles are ‘parameterized’: that is, they permit a finite set of options,
which individual languages draw on and which thus define how languages
differ. A language learner needs to discover which parameter settings apply in
the target language. See also principle and Parameter-setting Model.

Parameter-setting Model

This is a theory of L2 acquisition proposed by Flynn (1984). It is based on the

theory of Universal Grammar and assumes that adult L2 learners have con¬
tinued access to this. Their task is to discover how each principle is realized in
the L2 (i.e. what parameter settings are needed).

pattern of development

This term is used in this book to refer to the overall shape of L2 acquisition,
from the initial stages to the final stages. It serves as a cover term, therefore,
for both order of development and sequence of development.

patterns

Patterns are one type of formulaic speech. They are unanalysed units with one
or more open slots, for example, ‘Can I have a_?’. See also formulas and
routines.

performance

This term refers to the actual use of language in either comprehension or pro¬
duction. It contrasts with competence. See also linguistic competence.
f

pied piping

This is a term used in grammar. It refers to the proximity of two tied gram¬
matical elements. For example, the preposition ‘for’ and the pronoun ‘whom’
are placed next to each other in the sentence: ‘For whom did you get the pre¬
sent?’. Pied piping contrasts with preposition stranding.

planned discourse

Ochs (1979: 55) distinguishes ‘discourse that lacks forethought and organ¬
izational preparation’ (i.e. unplanned discourse) and ‘discourse that has been
thought out and organized prior to its expression’ (i.e. planned discourse).
See also speech planning.
 Glossary 719

positive transfer

According to behaviourist accounts of L2 acquisition, learners will have no

difficulty in learning L2 patterns when these are the same as LI patterns. Sup¬
port for positive transfer occurs when it can be shown that learners acquire
L2 features which are the same or similar to LI features with little difficulty.

poverty of the stimulus

Researchers in the Chomskyan tradition view the input available to learners

as an inadequate source of information for building a grammar. The stimulus
can be considered impoverished on several grounds: (1) input is degenerate
(for example, it contains ungrammatical sentences), (2) input underdeter¬
mines the grammar that needs to be constructed, and (3) the input to children
does not typically contain negative evidence.

pragmalinguistic failure

This is a term used by Thomas (1983). It arises when a learner tries to perform
the right speech act but uses the wrong linguistic means.

pragmatic competence

Pragmatic competence consists of the knowledge that speaker-hearers use in

order to engage in communication, including how speech acts are successfully
performed. Pragmatic competence is normally distinguished from linguistic
competence. Both are seen as relating to ‘knowledge’, and are therefore dis¬
tinct from actual performance.

pragmatics

Pragmatics is the study of how language is used in communication. Among

other aspects of language use, it includes the study of illocutionary acts.

preposition stranding

In some languages, like English, a preposition does not always have to be

proximate to the noun or pronoun to which it grammatically belongs, as in
this sentence: ‘Who did you get the present for?’. Preposition stranding con¬
trasts with pied piping.

principle

This is a term used in Chomsky’s theory of Universal Grammar. It refers to

highly abstract properties of grammar which underlie the rules of specific lan¬
guages. Principles are thought to constrain the form that grammatical rules
can take. They constitute part of a child’s innate knowledge of language.
720 Glossary

private speech

Some L2 learners who go through a silent period engage in private conversa¬

tions with themselves, thus, perhaps, preparing themselves for social speech
later.

proactive inhibition

Proactive inhibition is the way in which previous learning prevents or inhibits

the learning of new habits. L2 learners are hypothesized to experience diffi¬
culty in learning target-language forms that are different from first-language
forms. See interference and transfer.

procedural knowledge

Two related but different uses of procedural knowledge can be found in SLA
research. On the one hand, it is used in contrast to declarative knowledge to
refer to knowledge that has become proceduralized so that it is available for
automatic and unconscious use. On the other hand, it refers to knowledge of
the various strategies that learners employ to make effective use of their L2
knowledge in communication (for example, communication strategies).

production strategy

Production strategies refer to the utilization of linguistic knowledge in com¬

munication. They differ from communication strategies in that they do not
imply any communication problem and in that they are generally used with¬
out conscious awareness.

proficiency

L2 proficiency refers to a learner’s skill in using the L2. It can be contrasted

with the term ‘competence’. Whereas, competence refers to the knowledge of
the L2 a learner has internalized, proficiency refers to the learner’s ability to
use this knowledge in different tasks.

projection capacity

Zobl (1984) suggests that L2 learners (like LI learners) have the ability to
‘project’ knowledge of one rule to enable them to acquire another, implicated
rule, for which they have received no direct evidence in the input. Thus ac¬
quiring rule x automatically enables learners to acquire rule y, if x implicates
y-

prototypically

This term is used by Kellerman (1977) to refer to perceptions that learners

 Glossary 721

have regarding the structure of their own language. These perceptions lead
them to treat some structures as transferable and others as non-transferable.

pseudo-question

See open question.

psycbolinguistic context

This is a term used in this book to refer to the extent to which a particular con¬
text of use affords time for planning linguistic production and also whether it
encourages or discourages monitoring of output.

psychological distance

A term used by Schumann (1978a) in his Acculturation Model. It refers to the

distance between a learner and the target-language community resulting from
various psychological factors such as language shock and rigidity of ego
boundaries.

psychotypology

Kellerman (1978) has suggested that learners have perceptions regarding the
distance between their LI and the L2 they are trying to learn. These percep¬
tions constitute their psychotypology.

recreation continuum

Corder (1978a) has suggested that one way of viewing the interlanguage con¬
tinuum is as a recreation continuum. One possible starting point is ‘some ba¬
sic simple grammar’, which is recalled from an early stage of LI acquisition.
Learners gradually comlexify this system.

reference group

L2 learners—particularly those in natural settings—often have a choice of

language variety as the target model. Their choices are influenced by the par¬
ticular social group or groups they wish to identify with.

referential question

A referential question is a question that is genuinely information-seeking (for

example, ‘Why didn’t you do your homework?’). See also open question and
closed question.

resilient features

Resilient linguistic features are those that are acquired relatively easily even
Ill Glossary

when the only input available to the learner is deficient. Examples are word
order rules. Resilient features contrast with fragile features.

restrictive simplification

Researchers in the ZISA project (for example, Meisel 1980) used the term ‘re¬
strictive simplication’ to refer to learners’ continued use of simplified struc¬
tures such as the delection of function words, even though they have
developed knowledge of corresponding non-simplified structures. It con¬
trasts with elaborative simplification.

restructuring

This is a term used in information-processing theories of LI acquisition to re¬

fer to the qualitative changes that take place in learners’ interlanguage at cer¬
tain stages of development (see McLaughlin 1990a). For example, learners
may begin by representing past-tense forms as separate items and then shift to
representing them by means of a general rule for past-tense formation.

restructuring continuum

Corder (1978a) suggested that one way of viewing the interlanguage con¬
tinuum is as a restructuring continuum. The starting point is the learner’s LI.
The learner gradually replaces LI-based rules with L2-based rules.

routines

Routines are one type of formulaic speech. They are units that are totally
unanalysed and which are learnt as wholes. A common routine is ‘I don’t
know’. See also formulaic speech and patterns.

selective attention hypothesis

According to this hypothesis, formal instruction aids L2 acquisition not by

actually causing new linguistic features to become part of the learner’s in¬
terlanguage, but by providing the learner with ‘hooks, points of access’
(Lightbown 1985). In particular, it has been suggested that formal instruction
initiates the process of acquiring a feature by helping learners to notice a new
feature in the input.

semilingualism

In some contexts, when learners of an L2 develop negative attitudes towards

both their own culture and that of the target language, semilingualism can
result, i.e. the learners fail to develop full proficiency in either language. This
idea is controversial, however.
 Glossary 723

sequence of development

This book distinguishes order and sequence of development. ‘Sequence of de¬

velopment’ refers to the stages of acquisition through which a learner passes
in acquiring specific grammatical structures such as interrogatives, negatives,
and relative clauses. See also order of development.

silent period

Some L2 learners, especially children, undergo an often lengthy period during

which they do not try to speak in the L2. However, they may be learning the
L2 through listening to others speak it and may also be engaging in private
speech.

situational context

See context.

situation-specific anxiety

See anxiety.

skill-building hypothesis

According to Krashen (1989), the skill-building hypothesis states that rules

are first learnt consciously and then gradually automatized through practice.

social/affective strategies

These are one type of learning strategy. They concern the ways in which
learners elect to interact with other learners and native speakers (for example,
‘asking a teacher for repetition’—Chamot 1987).

social distance

This is a term used by Schumann (1978a) to account for why some L2 learn¬
ers learn very slowly or achieve low levels of proficiency. Various factors such
as the size of the learner’s L2 group and the learner’s desire to acculturate
influence the ‘distance’ between the learner and the target-language commun¬
ity.

Socio-educational Model of L2 Learning

This is a model of L2 learning developed by Gardner (1985). It posits that the

social and cultural milieu in which learners grow up determines the attitudes
and motivational orientation they hold towards the target language, its
speakers, and its culture. These in turn influence learning outcomes.
724 Glossary

sociolinguistic variables

A sociolinguistic variable is a linguistic feature that varies in accordance with

factors such as age, sex, social class, and ethnic membership. For example,
Labov (1972) illustrates how variation in/r/ in the speech of New Yorkers re¬
flects the social class of a speaker.

sociopragmatic failure

This is a term used by Thomas (1983). It occurs when a learner performs the
wrong illocutionary act for the situation and constitutes a deviation with re¬
gard to appropriateness of meaning. An example is a learner who apologizes
where a native speaker would thank someone.

Speech Accommodation Theory

Speech Accommodation Theory is a social-psychological model of language

use proposed by Giles to account for the dynamic nature of variation within
the course of a conversation. Speakers can converge (i.e. make their speech
similar to the style of their addressee) or diverge (i.e. make their speech differ¬
ent from the style of their addressee). In some situations speech maintenance
occurs (i.e. speakers make no changes). Speech accommodation is motivated
by the attitudes speakers hold towards their audience.

speech act

A speech act is an utterance that performs a locutionary and an illocutionary

meaning in communication. For example, ‘I like your dress’ is a speech act
concerning a proposition about a person’s dress with the illocutionary force
of a compliment.

speech planning

In some contexts of language use, speakers have the opportunity to plan their
speech, while in others they have to use the language more spontaneously.
Speech planning influences the choice of linguistic form. For example, L2
learners may use a target-language form in planned language use but an in¬
terlanguage form in unplanned language use. Speech planning is sometimes
investigated by studying ‘temporal variables’ (for example, speech rate and
pause length).

state anxiety

See anxiety.

structural and semantic simplification

Learner language produced by beginners is characteristically simpler than the

target language, and in many respects resembles a pidgin language. Structural
 Glossary 72 5

simplication is evident in the omission of grammatical functors. Semantic

simplification is evident in the omission of propositional elements (for ex¬
ample, ‘"'Hitting’ instead of ‘Kurt is hitting his sister’).

style shifting

Both native speakers and L2 learners use different variants of a variable form
depending on the degree of attention they pay to their speech (i.e. whether
they are accessing their vernacular or careful style). Labov refers to these
changes in speech as ‘style shifting’.

stylistic continuum

Tarone (1983), drawing on the work of Labov, suggests that learners intern¬
alize different ‘styles’ of language, ranging from a careful to a vernacular
style. The stylistic continuum accounts for variability in learner language.

submersion

The term submersion is used to refer to educational settings where L2 learners

are required to learn in classrooms where most of the students are native
speakers so that few input adjustments take place.

subtractive bilingualism

See bilingualism.

substratum transfer

See language transfer.

syntactization

See grammaticalization.

target-like use analysis

This is an extension of obligatory occasion analysis. It is designed to take into

account the incorrect use of specific grammatical features in contexts that do
not require them in the target language (for example, ‘"'Mary and Peter likes
travelling.’), as well as non-suppliance in contexts that require the feature (for
example, ‘"'Simon eat an apple every day.’).

task-induced variation

This is a blanket term used to refer to the variability in language use evident
when learners are asked to perform different tasks. Ultimately, it is traceable
to other sources (such as linguistic and situational contexts).
726 Glossary

teachability hypothesis

According to Pienemann (1985: 37) ‘the teachability hypothesis predicts that

instruction can only promote acquisition if the interlanguage is close to the
point when the structure to be taught is acquired in the natural setting.’ This
hypothesis is derived from the Multidimensional Model.

teacher talk

Teachers address classroom language learners differently from the way they
address other kinds of classroom learners. They make adjustments to both
language form and language function in order to facilitate communication.
These adjustments are referred to as ‘teacher talk’. See also foreigner talk.

think-aloud tasks

These are tasks designed to collect introspective data on the strategies learn¬
ers use. Learners are asked to perform a task (for example, completing a cloze
test) and to concurrently report the thought processes they are using to ac¬
complish the task.

trait anxiety

See anxiety.

transfer errors

These are errors in learner language that can be accounted for in terms of dif¬
ferences between the structures of the LI and the L2. See language transfer
and proactive inhibition.

transitional constructions

Dulay, Burt, and Krashen (1982) define transitional constructions as ‘the in¬
terim language forms that learners use while they are still learning the gram¬
mar,,of a language’. For example, before learners master the rule for English
negatives, they operate with interim rules (such as ‘no’ + verb).

tutor talk

See interlanguage talk,

typological universals

Typological universals are identified by examining a representative sample of

natural languages in order to identify features that are common to all or most
of these languages. Typological universals can be absolute (i.e. occur in all
languages), tendencies (i.e. occur in a large number of, but not all, languages),
 Glossary I'll

or implicational (i.e. the presence of one feature implies the presence of

another).

Universal Grammar

This is a term used by Chomsky to refer to the abstract knowledge of lan¬

guage which children bring to the task of learning their native language, and
which constrains the shape of the particular grammar they are trying to learn.
Universal Grammar consists of various principles which govern the form
grammatical rules can take. Some of these principles are ‘paramaterized’ (i.e.
are specified as consisting of two or more options). See also principles and
parameters.

unplanned discourse

See speech planning and planned discourse,

usage

Widdowson (1978:3) uses the term ‘usage’ to refer to ‘that aspect of perform¬
ance which makes evident the extent to which the language user demonstrates
his knowledge of linguistic rules’. The term contrasts with use.

use

Widdowson (1978: 3) uses the term ‘use’ to refer to that aspect of perform¬
ance which ‘makes evident the extent to which the language user demon¬
strates his ability to use his knowledge of linguistic rules for effective
communication’.

U-shaped behaviour

L2 learners have been observed to manifest a target-language form in their

output at an early stage of development only to manifest an interlanguage
form in its place at a later stage. Eventually the correct target-language form
reappears (for example, ‘came’ becomes ‘corned’ and, later still, ‘came’
again). This pattern of development is known as ‘U-shaped behaviour’.

variability hypothesis

This is a term used in this book to refer to the possibility that formal instruc¬
tion may have an effect on the learner’s careful style but is less likely to have
an effect on the learner’s vernacular style. See also stylistic continuum.

Variable Competence Model

A variable competence model seeks to account for the variability evident in

learner language by positing that this reflects a competence that is itself
728 Glossary

variable (i.e. it contains variable rules or different styles). Tarone (1983) has
proposed that the learner’s competence comprises a stylistic continuum.

variable form

A variable form is a phonological, lexical, or grammatical feature that is real¬

ized linguistically in more than one way. The linguistic devices that realize a
variable form are known as variants.

variants

A variable form has two or more variants, i.e. it can be realized by two or
more linguistic structures. For example, English copula has two vari¬
ants—contracted and full copula.

variational features

See Multidimensional Model.

vernacular style

This is a term used by Labov (1970) to refer to the language forms evident
when speakers are communicating spontaneously and easily with interloc¬
utors familiar to them. It contrasts with careful style, and for this reason is
also sometimes referred to as ‘casual style’. See also stylistic continuum.

vertical constructions

Vertical constructions are learner utterances which are formed by borrowing

chunks from the preceding discourse and then adding to these from the learn¬
er’s own resources. For example, a learner utterance like ‘No come here’
could be constructed by taking ‘come here’ from a previous utterance and
adding ‘no’. According to one view of L2 acquisition, vertical constructions
are the precursors of horizontal constructions.
r
vertical variation

This refers to the differences in learner language evident from one time to
another. It reflects the development that is taking place in the learner’s
interlanguage.

Wave Theory

This is a sociolinguistic theory developed by Bailey (1973) to account for

linguistic change. It explains how old rules spread from one speech
community to another and how new rules arise. It also accounts for how rules
spread from one linguistic environment to another.
 Glossary 729

wild grammar

Goodluck (1986) has used the term ‘wild grammar’ to refer to a grammar that
contains rules that contravene Universal Grammar. It is argued that children
do not in fact construct wild grammars.
730

Bibliography

Aboud, F. and R. Meade (eds.). 1974. Cultural Factors in Learning and Education. Bellingham,
Washington, 5th Western Washington Symposium on Learning.
Abbott, G. 1980. ‘Toward a more rigorous analysis of foreign language errors’. International
Review of Applied Linguistics 18: 121-34.
Abraham, R. 1983. ‘Relationships between use of the strategy of monitoring and cognitive style’.
Studies in Second Language Acquisition 6: 17-32.
Abraham, R. 1985. ‘Field independence-dependence and the teaching of grammar’. TESOL
Quarterly 20: 689-702.
Abraham, R. and R. Vann. 1987. ‘Strategies of two language learners: a case study’ in Wenden
and Rubin (eds.) 1987.
Adams, K. and D. Brinks (eds.). 1990. Perspectives on Official English: the Campaign for Eng¬
lish as the Official Language of the USA. Berlin: Mouton de Gruyter.
Adams, M. 1978. ‘Methodology for examining second language acquisition’ in Hatch (ed.)
1978.
Adamson, H. and V. Regan. 1991. ‘The acquisition of community norms by Asian immigrants
learning English as a second language: a preliminary study’. Studies in Second Language
Acquisition 1:1-22.
Adger, C. 1987. ‘Accommodating cultural differences in conversational style: a case study’ in
Lantolf and Labarca (eds.) 1987.
Adjemian, C. 1976. ‘On the nature of interlanguage systems’. Language Learning 26: 297-320.
Africa, H. 1980. ‘Language in education in a multilingual state’. LInpublished PhD thesis. Uni¬
versity of Toronto: Ontario Institute for Studies in Education.
Akagawa, Y. 1990. ‘Avoidance of relative clauses by Japanese high school students’. JACET
Kiyo 21.
Akagawa, Y. 1992. ‘A replication of Abraham’s study in “Field Dependence and the Teaching of
Grammar” ’. Unpublished paper, Tokyo: Temple University Japan.
Alamari, M. 1982. ‘Type and immediate effect of error correction for Hebrew second-language
learners’. Jerusalem: School of Education, Hebrew University. Cited in Cohen et al. 1991.
Albrechtsen, D., B. Henriksen, and C. Fterch. 1980. ‘Native speaker reactions to learners’
spoken interlanguage’. Language Learning 30: 365-96.
Allen, J. and S. P. Corder (eds.). 1974. The Edinburgh Course in Applied Linguistics, Vol. 3.
London: Oxford University Press.
Allen, J., M. Frohlich, and N. Spada. 1984. ‘The communicative orientation of language teach¬
ing: an observation scheme’ in Handscombe, Orem, and Taylor (eds.) 1984.
Allen, P., M. Swain, B. Harley, and J. Cummins. 1990. ‘Aspects of classroom treatment: toward
a more comprehensive view of second language education’ in Harley et al. (eds.) 1990.
Allwright, D. 1988. Observation in the Language Classroom. London: Longman.
Allwright, D. and K. Bailey. 1991. Focus on the Language Classroom: An Intrqduction to Class¬
room Research for Language Teachers. Cambridge: Cambridge University Press.
Allwright, R. 1975a. ‘Problems in the study of the language teacher’s treatment of learner error’
in Burt and Dulay (eds.) 1975.
Allwright, R. (ed.). 1975b. Working Papers: Language Teaching Classroom Research. Univer¬
sity of Essex, Department of Language and Linguistics.
Allwright, R. 1980. ‘Turns, topics and tasks: patterns of participation in language teaching and
learning’ in Larsen-Freeman (ed.) 1980.
Allwright, R. 1983. ‘Classroom-centred research on language teaching and learning: a brief his¬
torical overview’. TESOL Quarterly 17: 191-204.
 Bibliography 731

Allwright, R. 1984. ‘The importance of interaction in classroom language learning’. Applied

Linguistics 5:15 6-71.
Alpert, R. and R. Haber. 1960. ‘Anxiety in academic achievement situations’. Journal of Abnor¬
mal and Social Psychology 61: 207-15.
Altis, J. (ed.). 1968. Report of the Nineteenth Annual Round Table Meeting on Linguistics and
Language Studies. Georgetown University, Washington D.C.: Georgetown University
Press.
Altman, H. 1980. ‘Foreign language teaching: focus on the learner’ in Altman and Vaughan
James (eds.).
Altman, H. and C. Vaughan James (eds.). 1980. Foreign Language Teaching: Meeting Individual
Needs. Oxford: Pergamon.
Andersen, R. (ed.). 1979a. The Acquisition and Use of Spanish and English as First and Second
Languages. Washington D.C.: TESOL.
Andersen, R. 1979b. ‘Expanding Schumann’s Pidginization Hypothesis’. Language Learning
29: 105-19.
Andersen, R. 1980. ‘The role of creolization in Schumann’s Pidginization Hypothesis for second
language acquisition’ in Scarcella and Krashen (eds.) 1980.
Andersen, R. (ed.). 1981. New Dimensions in Second Language Acquisition Research. Rowley,
Mass.: Newbury House.
Andersen, R. 1983a. ‘Transfer to somewhere’ in Gass and Selinker (eds.) 1983.
Andersen, R. 1983b. ‘Introduction: a language acquisition interpretation of pidginization and
creolization’ in Andersen (ed.) 1983c.
Andersen, R. (ed.). 1983c. Pidginization and Creolization as Language Acquisition. Rowley,
Mass.: Newbury House.
Andersen, R. 1984a. ‘The One-to-One Principle of interlanguage construction’. Language
Learning 34: 77-95.
Andersen, R. (ed.). 1984b. Second Language: a Crosslinguistic Perspective. Rowley, Mass.:
Newbury House.
Andersen, R. 1990. ‘Models, processes, principles and strategies: second language acquisition in¬
side and outside of the classroom’ in VanPatten and Lee (eds.) 1990.
Anderson, J. 1976. Language, Memory, and Thought. Hillsdale, N.J.: Lawrence Erlbaum.
Anderson, J. 1980. Cognitive Psychology and its Implications. San Francisco: Freeman. (2nd ed.
1985).
Anderson, J. 1983. The Architecture of Cognition. Cambridge, Mass.: Harvard University Press.
d’Anglejan, A. 1978. ‘Language learning in and out of classrooms’ in Richards (ed.) 1978.
d’Anglejan, A., G. Painchaud, and C. Renaud. 1986. ‘Beyond the classroom: a study of commun¬
icative abilities in adult immigrants following intensive instruction’. TESOL Quarterly 20:
185-205.
d’Anglejan, A. and C. Renaud. 1985. ‘Learner characteristics and second language acquisition: a
multivariate study of adult immigrants and some thoughts on methodology’. Language
Learning 35:1-19.
Arabski, J. 1979. Errors as indicators of the development of interlanguage. Katowice: Universy-
tet Slaski.
Arnaud, P. and H. Bejoint (eds.). 1992. Vocabulary and Applied Linguistics. Basingstoke:
Macmillan.
Arthur, B. 1980. ‘Guaging the boundaries of second language competence: a study of learner
judgments’. Language Learning 30:177-94.
Arthur, B., M. Weiner, J. Culver, L. Young, and D. Thomas. 1980. ‘The register of impersonal
discourse to foreigners: verbal adjustments to foreign accent’ in Larsen-Freeman (ed.) 1980.
Asher, J. 1977. Learning Another Language Through Actions: The Complete Teachers’ Guide¬
book. Los Gatos, Calif.: Sky Oaks Publications.
732 Bibliography

Asher, J. and R. Garcia. 1969. ‘The optimal age to learn a foreign language’. Modern Language
Journal 53: 334-41.
Asher, J., J. Kusudo, and R. de la Torre. 1974. ‘Learning a second language through commands:
the second field test’. Modern Language Journal 58: 24-32.
Aston, G. 1986. ‘Trouble-shooting in interaction with learners: the more the merrier?’ Applied
Linguistics 7: 128-43.
Atkinson, M. 1986. ‘Learnability’ in Fletcher and Garman (eds.) 1986.
Au, S. 1988. ‘A critical appraisal of Gardner’s social-psychological theory of second language
(L2) learning’. Language Learning 38: 75-100.
Austin, J. 1962. How To Do Things With Words. Oxford: Clarendon Press.

Bach, E. and R. Harms (eds.). 1968. Universals of Linguistic Theory. New York: Holt, Rinehart,
and Winston.
Bachman, J. 1964. ‘Motivation in a task situation as a function of ability and control over task’.
Journal of Abnormal and Social Psychology 69: 272-81.
Bachman, L. 1990. Fundamental Considerations in Language Testing. Oxford: Oxford Univer¬
sity Press.
Bacon, S. 1987. ‘Differentiated cognitive style and oral performance’ in VanPatten et al. (eds.)
1987.
Bacon, S. 1992. ‘The relationship between gender, comprehension, processing strategies, and
cognitive and affective response in second-language listening’. Modern Language Journal
76: 160-78.
Bacon, S. and M. Finnemann. 1992. ‘Sex differences in self-reported beliefs about foreign-
language learning and authentic oral and written input’. Language Learning 42: 471-95.
Bahns, J. and H. Wode. 1980. ‘Form and function in L2 acquisition’ in Felix (ed.) 1980a.
Bailey, C. 1973. Variation and Linguistic Theory. Washington D.C.: Center for Applied
Linguistics.
Bailey, K. 1980. ‘An introspective analysis of an individual’s language learning experience’ in
Scarcella and Krashen (eds.) 1980.
Bailey, K. 1983. ‘Competitiveness and anxiety in adult second language learning: looking at and
through the diary studies’ in Seliger and Long (eds.) 1983.
Bailey, K. 1991. ‘Diary studies of classroom language learning: the doubting game and the be¬
lieving game’ in Sadtano (ed.) 1991.
Bailey, N. 1989. ‘Theoretical implications of the acquisition of the English simple past and past
progressive: putting together the pieces of the puzzle’ in Gass et al. (eds.) 1989b.
Bailey, K., M. Long, and S. Peak (eds.). 1983. Second Language Acquisition Studies. Rowley,
Mass: Newbury House.
Bailey, N., C. Madden, and S. Krashen. 1974. ‘Is there a “natural sequence” in adult second lan¬
guage learning?’ Language Learning 21: 235-43.
Baker, C. 1988. Key Issues in Bilingualism and Bilingual Education. Clevedon, Avon: Multi¬
lingual Matters.
Ball, P., H. Giles, and M. Hewstone. 1984. ‘The intergroup theory of second language acquisi¬
tion with catastrophic dimensions’ in Tajfel (ed.) 1984.
Banbrook, L. 1987. ‘Questions about questions: an inquiry into the study of teachers’ ques¬
tioning behaviour in ESL classrooms’. TESOL Quarterly 20: 47-59.
Banbrook, L. and P. Skehan. 1990. ‘Classroom and display questions’ in Brumfit and Mitchell
(eds.) 1990.
Bardovi-Harlig, K. 1987. ‘Markedness and salience in second language acquisition’. Language
Learning 37: 385^407.
Bardovi-Harlig, K. and T. Bofman. 1989. ‘Attainment of syntactic and morphological accuracy
by advanced language learners’. Studies in Second Language Acquisition 11: 17-34.
Bardovi-Harlig, K. and B. Hartford. 1990. ‘Congruence in native and nonnative conversations:
status balance in the academic advising session’. Language Learning 40: 467-501.
 Bibliography 733

Barnes, D. 1969. ‘Language in the secondary classroom’ in Barnes et al. (eds.) 1969.
Barnes, D. 1976. From Communication to Curriculum. Harmondsworth: Penguin.
Barnes, D., J. Britton, and M. Torbe (eds.). 1969. Language, the Learner and the School. Har¬
mondsworth: Penguin.
Barnes, S., M. Guttfreund, D. Satterly, and G. Wells. 1983. ‘Characteristics of adult speech
which predict children’s language development’. Journal of Child Language 10: 65-84.
Bates, E. and B. MacWhinney. 1982. ‘Functionalist approaches to grammar’ in Wanner and
Gleitman (eds.) 1982.
Bates, E. and B. MacWhinney. 1987. ‘Competition, variation and language learning’ in Mac-
Whinney (ed.) 1987.
Beck, M. and L. Eubank. 1991. ‘Acquisition theory and experimental design: a critique of Toma-
sello and Herron’. Studies in Second Language Acquisition 13: 73-6.
Beebe, L. 1974. ‘Socially conditioned variation in Bangkok Thai’. PhD dissertation, Ann Arbor:
University of Michigan.
Beebe, L. 1977. ‘The influence of the listener on code-switching’. Language Learning 27: 331-9.
Beebe, L. 1980. ‘Sociolinguistic variation and style-shifting in second language acquisition’.
Language Learning 30: 433^17.
Beebe, L. 1981. ‘Social and situational factors affecting the communicative strategy of dialect
code-switching’. International Journal of Sociology of Language 32: 139-49.
Beebe, L. 1982. ‘Reservations about the Labovian paradigm of style shifting and its extension to
the study of interlanguage’. Plenary paper presented at Los Angeles Second Language Re¬
search Forum.
Beebe, L. 1985. ‘Input: Choosing the right stuff’ in Gass and Madden (eds.) 1985.
Beebe, L. (ed.) 1988a. Issues in Second Language Acquisition: Multiple Perspectives. New York:
Newbury House.
Beebe, L. 1988b. ‘Five sociolinguistic approaches to second language acquisition’ in Beebe (ed.)
1988a.
Beebe, L. and M. Cummings. 1985. ‘Speech act performance: A function of the data collection
procedure’. Paper presented at TESOL Convention, New York.
Beebe, L. and M. Cummings, (forthcoming). ‘Natural speech act data vs. written questionnaire
data: how data collection method affects speech act performance’ in Neu and Gass (eds.).
Beebe, L. and H. Giles. 1984. ‘Speech accommodation theories: a discussion in terms of second
language acquisition’. International Journal of the Sociology of Language 46: 5-32.
Beebe, L. andT. Takahashi. 1989a. ‘ “Do you have a bag?” Social status and patterned variation
in second language acquisition’ in Gass et al. (eds.) 1989a.
Beebe, L. and T. Takahashi. 1989b. ‘Sociolinguistic variation in face-threatening speech acts’ in
Eisenstein (ed.) 1989.
Beebe, L., T. Takahashi, and R. Uliss-Weltz. 1990. ‘Pragmatic transfer in ESL refusals’ in Scar-
cella et al. (eds.) 1990.
Beebe, L. and J. Zuengler. 1983. ‘Accommodation theory: an explanation for style shifting in se¬
cond language dialects’ in Wolfson and Judd (eds.) 1983.
Bell, A. 1984. ‘Language style as audience design’. Language in Society 13: 145-204.
Bell, R. 1974. ‘Error analysis: a recent pseudoprocedure in applied linguistics’. ITL Review of
Applied Linguistics 25-6: 35-9.
Bellack, A., A. Herbert, M. Kliebard, R. Hyman, and F. Smith. 1966. The Language of the Class¬
room. New York: Teachers College Press.
Bennet-Kastor,T. 1988. Analyzing Children’s Language: Methods and Theories. Oxford:
Blackwell.
Berent, G. 1985. ‘Markedness considerations in the acquisition of conditional sentences’. Lan¬
guage Learning 35: 337-72.
Beretta, A. 1989. ‘Attention to form or meaning? Error treatment in the Bangalore Project’.
TESOL Quarterly 23: 283-303.
734 Bibliography

Beretta, A. 1991. ‘Theory construction in SLA: complementarity and opposition’. Studies in Se¬
cond Language Acquisition 13: 493-511.
Beretta, A. 1993. ‘ “As God said, and I think, rightly ..Perspectives on theory construction in
SLA: an introduction’. Applied Linguistics 14: 221-24.
Beretta, A. and A. Davies. 1985. ‘Evaluation of the Bangalore Project’. English Language Teach¬
ing Journal 39: 121-7.
Berko, J. 1958. ‘The child’s learning of English morphology’. Word 14: 150-77.
Berwick, R. 1985. The Acquisition of Syntactic Knowledge. Cambridge, Mass.: MIT Press.
Berwick, R. 1990. Task Variation and Repair in English as a Foreign Language. Kobe University
of Commerce: Institute of Economic Research.
Berwick, R. and S. Ross. 1989. ‘Motivation after matriculation: are Japanese learners of English
still alive after examination hell’. JALT11: 193-210.
Berwick, R. and A. Weinberg. 1984. The Grammatical Basis of Linguistic Performance: Lan¬
guage Use and Acquisition. Cambridge, Mass.: MIT Press.
Bhardwaj, M. 1986. ‘Reference to space by a Punjabi acquirer of English’. Nijmegen: ESF Work¬
ing paper.
Bhardwaj, M., R. Dietrich, and C. Noyan. 1988. ‘Second language acquisition by adult immig¬
rants: temporality’. Final Report Volume 5. Strasbourg, France: European Science
Foundation.
Bhatia, T. and W. Ritchie (eds.). (forthcoming). Handbook of Language Acquisition. New
York: Academic Press.
Bialystok, E. 1978. ‘A theoretical model of second language learning’. Language Learning 28:
69-84.
Bialystok, E. 1979. ‘Explicit and implicit judgements of L2 grammaticality’. Language Learning
29:81-104.
Bialystok, E. 1981a. ‘The role of linguistic knowledge in second language use’. Studies in Second
Language Acquisition 4: 31-45.
Bialystok, E. 1981b. ‘The role of conscious strategies in second language proficiency’. Modern
Language Journal 65: 24-35.
Bialystok, E. 1982. ‘On the relationship between knowing and using forms’. Applied Linguistics
3:181-206.
Bialystok, E. 1983a. ‘Some factors in the selection and implementation of communication strate¬
gies’ in Fterch and Kasper (eds.).
Bialystok, E. 1983b. ‘Inferencing: testing the “hypothesis-testing” hypothesis’ in Seliger and
Long (eds.) 1983.
Bialystok, E. 1985. ‘The compatibility of teaching and learning strategies’. Applied Linguistics 6:
155-262.
Bialystok, E. 1990a. Communication Strategies: A Psychological Analysis of Second-Language
fUse. Oxford: Basil Blackwell.
Bialystok, E. 1990b. ‘The competence of processing: classifying theories of second language ac¬
quisition’. TESOL Quarterly 24: 635-48.
Bialystok, E. 1991. ‘Achieving proficiency in a second language: a processing description’ in Phil-
lipson et al. (eds.) 1991.
Bialystok, E. and M. Frohlich. 1978. ‘Variables of classroom achievement in second language
learning’. Modern Language Journal 62: 327-36.
Bialystok, E., M. Frohlich, and J. Floward. 1978. ‘The teaching and learning of French as a se¬
cond language in two distinct learning settings’. Project report. Toronto: Modern Language
Centre, Ontario Institute for Studies in Education.
Bialystok, E. and E. Kellerman. 1987. ‘Communication strategies in the classroom’ in Das (ed.)
1987b.
Bialystok, E. and E. Ryan. 1985. ‘A metacognitive framework for the development of first and
second language skills’ in Forrest-Pressley et al. (eds.) 1985.
 Bibliography 735

Bialystok, E. and M. Sharwood Smith. 1985. ‘Interlanguage is not a state of mind: an evaluation
of the construct for second language acquisition’. Applied Linguistics 6: 101-17.
Bickerton, D. 1975. Dynamics of a Creole System. Cambridge: Cambridge University Press.
Bickerton, D. 1981. ‘Discussion of “Two perspectives on pidginization as second language ac¬
quisition” ’ in Andersen (ed.) 1981.
Bickerton, D. and T. Givon. 1976. ‘Pidginization and syntactic change: from SXV and VSX to
SVX’ in Stever et al. (eds). 1976.
Billmyer, K. 1990. ‘ “I really like your lifestyle”: ESL learners learning how to compliment’. Penn
Working Papers in Educational Linguistics 6: 31—48.
Bingaman, J. 1990. ‘On the English Proficiency Act’ in Adams and Brinks (eds.) 1990.
Birdsong, D. 1989. Metalinguistic Performance and Interlanguage Competence. New York:
Springer.
Birdsong, D. 1992. ‘Ultimate attainment in second language acquisition’. Language 68: 706-55.
Blatchford, C. and J. Schachter (eds.). 1978. On TESOL ’78: EFL Policies, Programs, Practices.
Washington D.C.: TESOL.
Blau, E. 1982. ‘The effect of syntax on readability for ESL students in Puerto Rico’. TESOL
Quarterly 16: 517-28.
Bley-Vroman, R. 1983. ‘The comparative fallacy in interlanguage studies: the case of system-
aticity’. Language Learning 33: 1-17.
Bley-Vroman, R. 1988. ‘The fundamental character of foreign language learning’ in Rutherford
and Sharwood Smith (eds.) 1988.
Bley-Vroman, R. 1989. ‘The logical problem of second language learning’ in Gass and Schachter
(eds.) 1989.
Bley-Vroman, R. and C. Chaudron. 1990. ‘Second language processing of subordinate clauses
and anaphora—first language and universal influences: a review of Flynn’s research’.
Language Learning 40: 245-85.
Bley-Vroman, R., S. Felix, and G. Ioup. 1988. ‘The accessibility of Universal Grammar in adult
language learning’. Second Language Research 4: 1-32.
Bloom, L. 1970. Language Development: Form and Function in Emerging Grammars. Cam¬
bridge, Mass.: MIT Press.
Bloomfield, L. 1933. Language. New York: Holt, Rinehart, and Winston.
Blum-Kulka, S. 1991. ‘Interlanguage pragmatics: The case of requests’ in Phillipson et al. (eds.)
1991.
Blum-Kulka, S., J. House, and G. Kasper (eds.). 1989a. Cross-cultural Pragmatics: Requests and
Apologies. Norwood, N.J.: Ablex.
Blum-Kulka, S., J. House, and G. Kasper. 1989b. ‘Investigating cross-cultural pragmatics: an in¬
troductory overview’ in S. Blum-Kulka et al. (eds.) 1989a.
Blum-Kulka, S. and E. Olshtain. 1984. ‘Requests and apologies: a cross-cultural study of speech
act realization patterns (CCSARP)’. Applied Linguistics 5:196-213.
Blum-Kulka, S. and E. Olshtain. 1986. ‘Too many words: length of utterances and pragmatic
failur e’. Journal of Pragmatics 8: 47-61.
Bodman, J. and M. Eisenstein. 1988. ‘May God increase your bounty: the expression of grati¬
tude in English by native and non-native speakers’. Cross Currents 15: 1-21.
Bogen, J. and J. Woodward. 1988. ‘Saving the phenomena’. Philosophical Review 97: 303-52.
Cited in Gregg 1993.
Bohn, O. 1986. ‘Formulas, frame structures, and stereotypes in early syntactic development:
some new evidence from L2 acquisition’. Linguistics 24: 185-202.
Bohn, O. and J. Flege. 1992. ‘The production of new and similar vowels by adult German
learners of English’. Studies in Second Language Acquisition 14: 131-58.
Bolte, H. and W. Herrlitz (eds.). 1985. Kommunikation im Sprachunterricht. Utrecht: Instituut
‘Franzen’, University of Utrecht.
Bonikowska, M. 1988. ‘The choice of opting out’. Applied Linguistics 9: 169-81.
736 Bibliography

Born, W. (ed.). 1979. The Foreign Language Learner in Today’s Classroom Environment.
Montpelier, Vermont: Northeast Conference on the Teaching of Foreign Languages.
Bourhis, R. and H. Giles. 1977. ‘The language of intergroup distinctiveness’ in Giles (ed.) 1977.
Bowerman, M. 1985. ‘What shapes children’s grammar?’ in Slobin (ed.) 1985a.
Boyle, J. 1987. ‘Sex differences in listening vocabulary’. Language Learning 37: 273-84.
Braine, M. 1971. ‘The acquisition of language in infant and child’ in Reed (ed.) 1971.
Breen, M. 1985. ‘The social context for language learning—a neglected situation?’ Studies in
Second Language Acquisition 7:135-58.
Breen, M. 1987. ‘Learner contributions to task design’ in Candlin and Murphy (eds.) 1987.
Briere, E. 1978. ‘Variables affecting native Mexican children’s learning Spanish as a second lan¬
guage’. Language Learning 28: 159-74.
Brindley, G. 1991. ‘Learnability in the ESL classroom: a pilot study’. Paper presented at RELC
Regional Seminar on Language Acquisition and the Second/Foreign Language Classroom,
Singapore.
Brock, C. 1986. ‘The effects of referential questions on ESL classroom discourse’. TESOL Quar¬
terly 20: 47-8.
Broeder, P., G. Extra, and R. van Flout. 1989. ‘Vocabulary acquisition’. AILA Review 1989.
Amsterdam: Free University Press.
Broen, P. 1972. ‘The verbal environment of the language learning child’. American Speech and
Hearing Monographs, No. 17.
Brooks, N. 1960. Language and Language Learning. New York: Llarcourt Brace and World.
Brown A., J. Bransford, R. Ferrara, and J. Campione. 1983. ‘Learning, remembering and under¬
standing’ in Flavell and Markman (eds.) 1983.
Brown, C. 1985. ‘Requests for specific language input: differences between older and younger
adult language learners’ in Gass and Madden (eds.) 1985.
Brown, H. 1980. ‘The optimal distance model of second language acquisition’. TESOL Quar¬
terly 14:157-64.
Brown, H. 1987. Principles of Language Learning and Teaching (2nd Ed.). Englewood Cliffs,
N.J.: Prentice Flail.
Brown, H. 1989. A Practical Guide to Language Learning. New York: McGraw FLill.
Brown, FL, C. Yorio, and R. Crymes (eds.). 1977. On TESOL ’77. Washington D.C.: TESOL.
Brown, P. and C. Fraser. 1979. ‘Speech as markers of situation’ in Scherer and Giles (eds.) 1979.
Brown, P. and S. Levinson. 1978. ‘Universals of language usage: politeness phenomena’ in
Goody (ed.) 1978.
Brown, R. 1968. ‘The development of Wh-questions in children’s speech’. Journal of Verbal
Learning and Language Behavior 7: 279-90.
Brown, R. 1973. A First Language: the Early Stages. Cambridge, Mass.: Harvard University
Press.
Brown, R. 1977. ‘Introduction’ in Snow and Ferguson (eds.) 1977.
Brown, R. 1987. ‘A comparison of the comprehensibility of modified and unmodified reading
materials for ESL’. University of Hawaii Working Papers in ESL 6: 49-79.
Brown, R. 1991. ‘Group work, task difference, and second language acquisition’. Applied Lin¬
guistics 21: 1-12.
Brown, R. and C. Hanlon. 1970. ‘Derivational complexity and order of acquisition in child
speech’ in Hayes (ed.) 1970.
Brown, T. and F. Perry. 1991. ‘A comparison of three learning strategies for ESL vocabulary ac¬
quisition’. TESOL Quarterly 25: 655-70.
Brumfit, C. 1984. Communicative Methodology in Language Teaching. Cambridge: Cambridge
University Press.
Brumfit, C. and R. Carter (eds.). 1986. Literature and Language Teaching. Oxford: Oxford Uni¬
versity Press.
 Bibliography 737

Brumfit, C. and R. Mitchell (eds.). 1990. ‘Research in the Language Classroom’. ELT Docu¬
ments 133: Modern English Publications.
Bruner, J., J. Goodnow, and G. Austin. 1957. A Study of Thinking. New York: Wiley and Sons.
Buczowska, E. and R. Weist. 1991. ‘The effects of formal instruction on the second-language
acquisition of temporal location’. Language Learning 41: 535-54.
Burmeister, R. and P. Rounds (eds.). 1990. Variability in Second Language Acquisition: Pro¬
ceedings of the Tenth Meeting of the Second Language Research Forum, Vol 1. Eugene,
Oregon: University of Oregon.
Burstall, C. 1975. ‘Factors affecting foreign-language learning: a consideration of some relevant
research findings’. Language Teaching and Linguistics Abstracts 8: 105-25.
Burstall, C. 1979. ‘Primary French in the balance’ in Pride (ed.) 1979.
Burt, M. 1975. ‘Error analysis in the adult EFL classroom’. TESOL Quarterly 9: 53-63.
Burt, M. and H. Dulay (eds.). 1975. On TESOL ’75: New Directions in Second Language
Learning, Teaching and Bilingual Education. Washington D.C.: TESOL.
Burt, M., H. Dulay, and M. Finnocchario (eds.). 1977. Viewpoints on English as a Second Lan¬
guage. New York: Regents.
Burt, M., H. Dulay, and E. Hernandez. 1973. Bilingual Syntax Measure. New York: Harcourt
Brace Jovanovich.
Burt, M. and C. Kiparsky. 1972. The Gooficon: a Repair Manual for English. Rowley, Mass.:
Newbury House.
Busch, D. 1982. ‘Introversion-extroversion and the EFL proficiency of Japanese students’. Lan¬
guage Learning 32: 109-32.
Bush, R., E. Galanter, and R. Luce (eds.). 1963. Handbook of Mathematical Psychology, Vol II.
New York: Wiley and Sons.
Butterworth, B. (ed.). 1980a. Language Production, Vol. 1. New York: Academic Press.
Butterworth, B. 1980b. ‘Some constraints on models of language production’ in Butterworth
(ed.) 1980a.
Butterworth, G. and E. Hatch. 1978. ‘A Spanish-speaking adolescent’s acquisition of English
syntax’ in Hatch (ed.) 1978a.
Bygate, M. 1988. ‘Units of oral expression and language learning in small group interaction’.
Applied Linguistics 9: 59-82.

Canale, M. 1983. ‘From communicative competence to language pedagogy’ in Richards and

Schmidt (eds.) 1983.
Canale, M. and M. Swain. 1980. ‘Theoretical bases of communicative approaches to second lan¬
guage teaching and testing’. Applied Linguistics 1: 1-47.
Cancino, H., E. Rosansky, and J. Schumann. 1978. ‘The acquisition of English negatives and
interrogatives by native Spanish speakers’ in Hatch (ed.) 1978.
Candlin, C. and D. Murphy (eds.). 1987. Lancaster Practical Papers in English Language Educa¬
tion: Vol 7. ‘Language Learning Tasks’. Englewood Cliffs, N.J.: Prentice Hall.
Carlisle, R. 1991. ‘The influence of environment on vowel epenthesis in Spanish/English in¬
terphonology’. Applied Linguistics 12: 76-95.
Carrell, P. 1981. ‘Relative difficulty of request forms in L1/L2 comprehension’ in Hines and
Rutherford (eds.) 1981.
Carrell, P. and B. Konneker. 1981. ‘Politeness: comparing native and nonnative judgments’.
Language Learning 31:17-31.
Carroll, J. 1965. ‘The prediction of success in foreign language training’ in Glaser (ed.) 1965.
Carroll, J. 1967. ‘Foreign language proficiency levels attained by language majors near gradu¬
ation from college’. Foreign Language Annals 1: 131-51.
Carroll, J. 1969. ‘Reaction to Brimer’s article’. Journal of Research and Development in Educa¬
tion 9:18-1.
738 Bibliography

Carroll, J. 1981. ‘Twenty-five years of research on foreign language aptitude’ in Diller (ed.)
1981.
Carroll, J. 1990. ‘Cognitive abilities in foreign language aptitude: then and now’ in Parry and
Stansfield (eds.) 1990.
Carroll, J. and S. Sapon. 1959. Modern Language Aptitude Test—Form A. New York: The Psy¬
chological Corporation.
Carroll, J. and D. Spearritt. 1967. ‘A study of “A Model of School Learning” ’. Unpublished re¬
port, Graduate School of Education, Harvard University. Cited in Cronbach and Snow
1977.
Carroll, S., M. Swain, and Y. Roberge. 1992. ‘The role of feedback in adult second language ac¬
quisition: error correction and morphological generalizations’. Applied Psycholinguistics
13:173-98.
Carter, E. 1988. ‘The relationship of field dependent/independent cognitive style to Spanish lan¬
guage achievement and proficiency: a preliminary report’. Modern Language Journal 72:
21-30.
Carton, A. 1971. ‘Inferencing: a process in using and learning language’ in Pimsleur and Quinn
(eds.) 1971.
Castagnaro, P. 1992. ‘Introduction of physiological measures in locating aversive stimulation re¬
lated to second language learning among Japanese university students’. Unpublished manu¬
script, Tokyo: Temple University Japan.
Cathcart, R. 1986. ‘Situational differences and the sampling of young children’s school lan¬
guage’ in Day (ed.) 1986.
Cathcart, R. and J. Olsen. 1976. ‘Teachers’ and students’ preferences for correction of classroom
errors’ in Fanselow and Crymes (eds.) 1976.
Cattell, R. 1956. ‘Personality and motivation theory based on structural measurement’ in
McCary (ed.) 1956.
Cazden, C. 1972. Child Language and Education. New York: Holt, Rinehart, and Winston.
Cazden, C., E. Cancino, E. Rosansky, and J. Schumann. 1975. Second Language Acquisition in
Children, Adolescents and Adults. Final Report. Washington D.C.: National Institute of
Education.
Cazden, C., V. John, and D. Hymes (eds.). 1972. Functions of Language in the Classroom. New
York: Teachers’ College Press.
Ceci, S. (ed.). 1987. Flandbook of Cognitive, Social and Neuropsychological Aspects of Learn¬
ing Disabilities. Hillsdale, N.J.: Lawrence Erlbaum.
Celce-Murcia, M. (ed.). 1985. Beyond Basics: Issues and Research in TESOL. Rowley, Mass.:
Newbury House.
Cervantes, R. 1983. ‘ “Say it again Sam”: the effect of repetition on dictation scores’. Term
paper, ESL 70, University of Hawaii at Manoa. Cited in Chaudron 1988.
Chanjbers, J. and P. Trudgill. 1980. Dialectology. Cambridge: Cambridge University Press.
Chamot, A. 1978. ‘Grammatical problems in learning English as a third language’ in Hatch (ed.)
1978.
Chamot, A. 1979. ‘Strategies in the acquisition of English structures by a child bilingual in Span¬
ish and French’ in Andersen (ed.) 1979a.
Chamot, A. 1987. ‘The learning strategies of ESL students’ in Wenden and Rubin (eds.) 1987.
Chamot, A., J. O’Malley, L. Kupper, and M. Impink-Hernandez. 1987. A Study of Learning
Strategies in Foreign Language Instruction: First year Report. Rosslyn, Va.: Interstate Re¬
search Associates.
Chamot, A., L. Kupper, and M. Impink-Hernandez. 1988. A Study of Learning Strategies in For¬
eign Language Instruction: Findings of the Longitudinal Study. McLean, Va.: Interstate Re¬
search Associates.
 Bibliography 739

Chapelle, C. 1992. ‘Disembedding “Disembedded Figures in the Landscape an appraisal of

Griffiths and Sheen’s “Reappraisal of L2 research on field dependence/independence”
Applied Linguistics 13: 375-84.
Chapelle, C. and P. Green. 1992. ‘Field independence/dependence in second language acquisi¬
tion research’. Language Learning 42: 47-83.
Chapelle, C. and C. Roberts. 1986. ‘Ambiguity tolerance and field independence as predictors of
proficiency in English as a second language’. Language Learning 36: 27-45.
Chastain, K. 1975. ‘Affective and ability factors in second language acquisition’. Language
Learning 25: 153-61.
Chastain, K. 1981. ‘Native-speaker evaluation of student composition errors’. Modern Lan¬
guage Journal 65: 288-94.
Chaudron, C. 1977. ‘A descriptive model of discourse in the corrective treatment of learners’
errors’. Language Learning 27: 29-46.
Chaudron, C. 1982. ‘Vocabulary elaboration in teachers’ speech to L2 learners’. Studies in
Second Language Acquisition 4: 170-80.
Chaudron, C. 1983a. ‘Research on metalinguistic judgements: a review of theory, methods and
results’. Language Learning 33: 343-77.
Chaudron, C. 1983b. ‘Foreigner talk in the classroom—an aid to learning?’ in Seliger and Long
(eds.) 1983.
Chaudron, C. 1985. ‘A method for examining the input/intake distinction’ in Gass and Madden
(eds.) 1985.
Chaudron, C. 1986. ‘Teachers’ priorities in correcting learning errors in French immersion
classes’ in Day (ed.) 1986.
Chaudron, C. 1987. ‘The role of error correction in second language teaching’ in Das (ed.)
1987a.
Chaudron, C. 1988. Second Language Classrooms: Research on Teaching and Learning. Cam¬
bridge: Cambridge University Press.
Chaudron, C. and J. Richards. 1986. ‘The effect of discourse markers on the comprehension of
lectures’. Applied Linguistics 7: 113-27.
Chenoweth, A., R. Day, A. Chun, and S. Luppescu. 1983. ‘Attitudes and preferences of nonnat¬
ive speakers to corrective feedback’. Studies in Second Language Acquisition 6: 79-87.
Chesterfield R. and K. Chesterfield. 1985. ‘Natural order in children’s use of second language
learning strategies’. Applied Linguistics 6: 45-59.
Chiang, D. 1980. ‘Predictors of relative clause production’ in Scarcellaand Krashen (eds.) 1980.
Chick, K. 1985. ‘The interactional accomplishment of discrimination in South Africa’. Language
in Society 14: 299-326.
Chihara, T. and J. Oiler. 1978. ‘Attitudes and attained proficiency in EFL: a sociolinguistic study
of adult Japanese speakers’. Language Learning 28: 55-68.
Chomsky, C. 1969. The Acquisition of Syntax in Children from 5 to 10. Cambridge, Mass.: MIT
Press.
Chomsky, N. 1959. ‘Review of “Verbal Behavior” by B. F. Skinner’. Language 35: 26-58.
Chomsky, N. 1965. Aspects of the Theory of Syntax. Cambridge, Mass.: MIT Press.
Chomsky, N. 1976. Reflections on Language. London: Temple Smith.
Chomsky, N. 1980a. Rules and Representations. New York: Columbia University Press.
Chomsky, N. 1980b. ‘On cognitive structures and their development: a reply to Piaget’ in Pia-
telli-Palmarini (ed.) 1980.
Chomsky, N. 1981a. ‘Principles and parameters in syntactic theory’ in Hornstein and Lightfoot
(eds.) 1981.
Chomsky, N. 1981b. Lectures on Government and Binding. Dordrecht: Foris.
Chomsky, N. 1986a. Knowledge of Language: its Nature, Origin and Use. New York: Praeger.
Chomsky, N. 1986b. Barriers. Cambridge, Mass.: MIT Press.
740 Bibliography

Chomsky, N. 1988. Language and Problems of Knowledge: the Nicaraguan Lectures. Cam¬
bridge, Mass.: MIT Press.
Chun, A., R. Day, A. Chenoweth, and S. Luppescu. 1982. ‘Errors, interaction, and correction: a
study of non-native conversations’. TESOL Quarterly 16: 537-47.
Cicourel, A. et al. (eds.). 1974. Language Use and School Performance. New York: Academic
Press.
Clahsen, H. 1980. ‘Psycholinguistic aspects of L2 acquisition’ in Felix (ed.) 1980a.
Clahsen, H. 1982. Spracherwerb in der Kindheit: eine Untersuchung zur Entwicklung der Syn¬
tax bei Kleinkindern. Tubingen: Gunter Narr.
Clahsen, H. 1984. ‘The acquisition of German word order: a test case for cognitive approaches
to L2 development’ in Andersen (ed.) 1984b.
Clahsen, H. 1988. ‘Critical phases of grammar development: a study of the acquisition of nega¬
tion in children and adults’ in Jordens and Talleman (eds.) 1988.
Clahsen, H. 1990. ‘The comparative study of first and second language development’. Studies in
Second Language Acquisition 12: 135-54.
Clahsen, H., J. Meisel, and M. Pienemann. 1983. Deutsch als Zweitsprache: der Spracherwerb
auslandischer Arbeiter. Gunter Narr: Tubingen.
Clahsen, H. and P. Muysken. 1986. ‘The availability of universal grammar to adult and child
learners—the study of the acquisition of German word order’. Second Language Research
2:93-119.
Clahsen, H. and P. Muysken. 1989. ‘The UG paradox in L2 acquisition’. Second Language Re¬
search 5: 1-29.
Clark, H. and E. Clark. 1977. Psychology and Language: an Introduction to Psycholinguistics.
New York: Harcourt Brace Jovanovich.
Clark J. 1969. ‘The Pennsylvania Project and the “Audio-Lingual vs. Traditional” question’.
Modern Language Journal 53: 388-96.
Clark, R. 1974. ‘Performing without competence’. Journal of Child Language 1:1-10.
Clarke, M. and J. Hanscombe (eds.). 1983. On TESOL ’82. Washington D.C.: TESOL.
Clement, R. 1980. ‘Ethnicity, contact and communicative competence in a second language’ in
Giles et al. (eds.) 1980.
Clement, R. 1986. ‘Second language proficiency and acculturation: an investigation of the effects
of language status and individual characteristics’. Journal of Language and Social Psychol¬
ogy 5:271-90.
Clement, R. and B. Kruidenier. 1983. ‘Orientations in second language acquisition: 1. The ef¬
fects of ethnicity, milieu and target language on their emergence’. Language Learning 33:
273-91.
Clement, R. and B. Kruidenier. 1985. ‘Aptitude, attitude and motivation in second language pro¬
ficiency: a test of Clement’s model’. Journal of Language and Social Psychology 4: 21-38.
Clement, R., P. Smythe, and R. Gardner. 1978. ‘Persistence in second language study: motiva¬
tional considerations’. Canadian Modern Language Review 34: 688-94.
Clyne, M. (ed.). 1975. ‘Foreigner Talk’. International Journal of the Sociology of Language 28.
Clyne, M. 1978. ‘Some remarks on foreigner talk’ in Dittmar et al. (eds.) 1978.
Coan, R. and R. Cattell. 1966. Early School Personality Questionnaire. Champaign, Illinois:
Institute for Personality and Ability Testing.
Cochrane, R. 1980. ‘The acquisition of Irl and/1/ by Japanese children and adults learning Eng¬
lish as a second language’. Journal of Multilingual and Multicultural Development 1:
331-60.
Cohen, A. 1984. ‘Studying second-language learning strategies: how do we get the information?’
Applied Linguistics 5: 101-12. Also in Wenden and Rubin (eds.) 1987.
Cohen, A. 1987. ‘Student processing of feedback on their compositions’ in Wenden and Rubin
(eds.) 1987.
 Bibliography 741

Cohen, A. 1990. Language Learning: Insights for Learners, Teachers, and Researchers. New
York: Newbury House/Harper Row.
Cohen, A. 1991. ‘Feedback on writing: the use of verbal report’. Studies in Second Language
Acquisition 13: 133-59.
Cohen, A. and E. Aphek. 1980. ‘Retention of second language vocabulary over time: in¬
vestigating the role of mnemonic associations’. System 8: 221-35.
Cohen, A. and E. Aphek. 1981. ‘Easifying second language learning’. Studies in Second Lan¬
guage Acquisition 3: 221-36.
Cohen, A., D. Larsen-Freeman, and E. Tarone. 1991. ‘The contribution of SLA theories and re¬
search to language teaching’ in Sadtano (ed.) 1991.
Cohen, A. and E. Olshtain. 1981. ‘Developing a measure of sociocultural competence; the case of
apology’. Language Learning 31: 113-34.
Cohen, A. and M. Swain. 1979. ‘Bilingual education: the “immersion” model in the North Am¬
erican context’ in Pride (ed.) 1979.
Cole, P. and J. Morgan (eds.). 1975. Syntax and Semantics 3: Speech Acts. New York: Academic
Press.
Coleman, L. and P. Kay. 1981. ‘Prototype semantics: the English word “lie” ’. Language 55:
26-44.
Collier, V. 1992. ‘The Canadian bilingual immersion debate: a synthesis of research findings’.
Studies in Second Language Acquisition 14: 87-97.
Committee on Irish Language Attitudes Research (CILAR). 1975. ‘Report of the Committee on
Irish Language Attitudes Research’. Dublin: Government Stationery Office. Cited in Baker
1988.
Comrie, B. 1984. Language Universal and Linguistic Typology. Oxford: Basil Blackwell.
Comrie, B. and E. Keenan. 1979. ‘Noun phrase accessibility revisited’. Language 55: 649-64.
Conrad, L. 1989. ‘The effects of time-compressed speech on native and EFL listening compre¬
hension’. Studies in Second language Acquisition 11: 1-16.
Cook, V. 1971. ‘The analogy between first and second language learning’ in Lugton (ed.) 1971.
Cook, V. 1973. ‘The comparison of language development in native children and foreign adults’.
International Review of Applied Linguistics 11: 13-28.
Cook, V. 1977. ‘Cognitive processes in second language learning’. International Review of
Applied Linguistics 15: 1-20.
Cook, V. 1985. ‘Chomsky’s universal grammar and second language learning’. Applied Lin¬
guistics 6: 2-18.
Cook, V. (ed.). 1986. Experimental Approaches to Second Language Acquisition. Oxford:
Pergamon.
Cook, V. 1988. Chomsky’s Universal Grammar: an Introduction. Oxford: Basil Blackwell.
Cook, V. 1989. ‘Universal grammar theory and the classroom’. System 17:169-82.
Cook, V. 1990. ‘Timed comprehension of binding in advanced L2 learners of English’. Language
Learning AO: 557-99.
Coppieters, R. 1987. ‘Competence differences between native and near-native speakers’. Lan¬
guage 63:544-73.
Corder, S. P. 1967. ‘The significance of learners’ errors’. International Review of Applied Lin¬
guistics 5: 161-9.
Corder, S. P. 1971a. ‘Idiosyncractic dialects and error analysis’. International Review of Applied
Linguistics 9: 149-59.
Corder, S. P. 1971b. ‘Describing the language learner’s language’. CILT Reports and Papers,
No. 6. CILT.
Corder, S. P. 1973. ‘The elicitation of interlanguage’ in Svartvik (ed.) 1973a.
Corder, S. P. 1974. ‘Error analysis’ in Allen and Corder (eds.) 1974.
Corder, S. P. 1976. ‘The study of interlanguage’ in Proceedings of the Fourth International Con¬
ference of Applied Linguistics. Munich, Hochschulverlag. Also in Corder 1981a.
742 Bibliography

Corder, S. P. 1977a. ‘ “Simple codes” and the source of the learner’s initial heuristic hypothesis’.
Studies in Second Language Acquisition 1: 1-10.
Corder, S. P. 1977b. ‘Language teaching and learning: a social encounter’ in Brown et al. (eds.)
1977.
Corder, S. P. 1978a. ‘Language-learner language’ in Richards (ed.) 1978.
Corder, S. P. 1978b. ‘Language distance and the magnitude of the learning task’. Studies in
Second Language Acquisition 2: 27-36.
Corder, S. P. 1978c. ‘Strategies of communication’. AFinLa 23. Also in Corder 1981a.
Corder, S. P. 1980. ‘Second language acquisition research and the teaching of grammar’. BAAL
Newsletter 10.
Corder, S. P. 1981a. Error Analysis and Interlanguage. Oxford: Oxford University Press.
Corder, S. P. 1981b. ‘Formal simplicity and functional simplification’ in Andersen (ed.) 1981.
Corder, S. P. 1983. ‘A role for the mother tongue’ in Gass and Selinker (eds.) 1983.
Corder, S. P. and H. Roulet (eds.). 1977. Actes du V erne colloque de Linguistt'que appliquee de
Neuchatel. Geneva: Droz et University de Neuchatel.
Coulmas, F. (ed.). 1981. Conversational Routines: Explorations in Standardized Communica¬
tion Situations and Prepatterned Speech. The Hague: Mouton.
Coulthard, M. 1985. An Introduction to Discourse Analysis. London: Longman.
Coulthard, M. and M. Montgomery (eds.). 1981. Studies in Discourse Analysis. Routledge,
Kegan, and Paul.
Coupland, N., H. Giles, and J. Weimann (eds.). 1991. Miscommunication and Problematic
Talk. Newbury Park: Sage Publications.
Courchene, R. 1980. ‘The error analysis hypothesis, the contrastive analysis hypothesis, and the
correction or error in the second language classroom’. TESL Talk 11/2: 3-13 and 11/3:
1-29.
Courchene, R., J. Glidden, J. St. John, and C. Therien (eds.). 1992. Comprehension-based
Second Language Teaching. Ottowa: University of Ottawa Press.
Crago, M. 1992. ‘Communicative interaction and second language acquisition: an Inuit ex¬
ample’. TESOL Quarterly 26: 487-505.
Craik, F. and R. Lockhart. 1972. ‘Levels of processing: a framework for memory research’, jour¬
nal of Verbal Learning and Verbal Behavior 11: 671-84.
Croft, W. 1990. Typology and Universals. Cambridge: Cambridge University Press.
Cronbach, L. and R. Snow. 1977. Aptitudes and Instructional Methods. New York: Irvington.
Crookes, G. 1986. ‘Task classification: a cross-disciplinary review’. Technical report No. 4.
Honolulu: Center for Second Language Classroom Research, Social Science Research Insti¬
tute, University of Hawaii.
Crookes, G. 1989. ‘Planning and interlanguage variability’. Studies in Second Language Acquisi¬
tion 11:367-83.
Crookes, G. 1991. ‘Second language speech production research: a methodologically oriented
review’. Studies in Second Language Acquisition 13: 113-32.
Crookes, G. 1992. ‘Theory format and SLA theory’. Studies in Second Language Acquisition 14:
425^19.
Crookes, G. and K. Rulon. 1985. ‘Incorporation of corrective feedback in native speaker/non¬
native speaker conversation’. Technical report No. 3. Honolulu: Center for Second Lan¬
guage Classroom Research, Social Science Research Institute, University of Hawaii.
Crookes, G. and R. Schmidt. 1989. ‘Motivation: Reopening the research agenda’. University of
Hawaii Working Papers in ESL 8: 217-56.
Cross, T. 1977. ‘Mother’s speech adjustments: the contribution of selected child listener vari¬
ables’ in Snow and Ferguson (eds.) 1977.
Cross, T. 1978. ‘Mothers’ speech and its association with rate of linguistic development in young
children’ in Waterson and Snow (eds.) 1978.
 Bibliography 743

Crystal, D. 1976. Child Language Learning and Linguistics: an Overview for the Teaching and
Therapeutic Professions. London: Edward Arnold.
Crystal, D. 1988. The English Language. Harmondsworth: Penguin.
Cummins, J. 1981. Bilingualism and Minority Children. Ontario: Ontario Institute for Studies in
Education.
Cummins, J. 1983. ‘Language proficiency and academic achievement’ in Oiler (ed.) 1983.
Cummins, J. 1984. Bilingualism and Special Education: Issues in Assessment and Pedagogy.
Clevedon, Avon: Multilingual Matters.
Cummins, J. 1988. ‘Second language acquisition within bilingual education programs’ in Beebe
(ed.) 1988a.
Cummins, J., B. Harley, M. Swain, and P. Allen. 1990. ‘Social and individual factors in the devel¬
opment of bilingual proficiency’ in Harley et al. (eds.) 1990.
Cummins, J. and K. Nakajima. 1987. ‘Age of arrival, length of residence, and interdependence of
literacy skills among Japanese immigrant students’ in Harley et al. (eds.) 1987.
Cummins, J. and M. Swain. 1986. Bilingualism in Education. London: Longman.
Cummins, J., M. Swain, K. Nakajima, J. Handscombe, D. Green, and C. Tran. 1984. ‘Linguistic
interdependence among Japanese and Vietnamese immigrant students’ in Rivera (ed.) 1984.

Dagut, M. and B. Laufer. 1985. ‘Avoidance of phrasal verbs—a case for contrastive analysis’.
Studies in Second Language Acquisition 7: 73-9.
Dahl, D. 1981. ‘The role of experience in speech modifications for second language learners’.
Minnesota Papers in Linguistics and Philosophy of Language 7: 78-93.
Dahl, O. 1979. ‘Typology of sentence negation’. Linguistics 17: 79-106.
Dakin, J. 1973. The Language Laboratory and Language Learning. London: Longman.
Daniel, I. 1983. ‘On first-year German foreign language learning: a comparison of language be¬
havior in response to two instructional methods’. Unpublished PhD thesis, University of
Southern California.
Danielewicz, J. 1984. ‘The interaction between text and context: a study of how adults and chil¬
dren use spoken and written language in four contexts’ in Pellegrini and Yawkey (eds.)
1984.
Danoff, M., G. Coles, D. McLaughlin, and D. Reynolds. 1978. ‘Evaluation of the impact of
ESEA Title VII Spanish/English bilingual education program: overview of study and End¬
ings’. Palo Alto, Calif.: American Institutes for Research.
Darden, L. 1991. Theory Change in Science: Strategies from Mendalian Genetics. New York:
Oxford University Press. Cited in Long 1993.
Das, B. (ed.). 1987a. Patterns of Classroom Interaction. Singapore: SEAMEO Regional Lan¬
guage Centre.
Das, B. (ed.). 1987b. Communication and Learning in the Classroom Community. Singapore:
SEAMEO Regional Language Centre.
Davies, A. 1989. ‘Is international English an interlanguage?’ TESOL Quarterly 23: 447-67.
Davies, A., C. Criper, and A. Howatt (eds.). 1984. Interlanguage. Edinburgh: Edinburgh Univer¬
sity Press.
Davies, E. 1983. ‘Error evaluation: the importance of viewpoint’. English Language Teaching
Journal 37: 304-11.
Day, R. 1984. ‘Student participation in the ESL classroom, or some imperfections of practice’.
Language Learning 34: 69-102.
Day, R. 1985. ‘The use of the target language in context and second language proficiency’ in Gass
and Madden (eds.) 1985.
Day, R. (ed.). 1986. Talking to Learn: Conversation in Second Language Acquisition. Rowley,
Mass.: Newbury House.
Day, R., N. Chenoweth, A. Chun, and S. Luppescu. 1984. ‘Corrective feedback in native-
nonnative discourse’. Language Learning 34:19-45.
744 Bibliography

de Beaugrande, R. and W. Dressier. 1981. Introduction to Text Linguistics. London: Longman.

de Bot, K. 1992. ‘A bilingual production model: Levelt’s “Speaking” model adapted’. Applied
Linguistics 13: 1-24.
de Bot, K., D. Coste, R. Ginsberg, and C. Kramsch (eds.). 1991. Foreign Language Research in
Cross-cultural Perspectives. Amsterdam: John Benjamins.
Decamp, D. 1971. ‘Implicational scales and sociolinguistic theory’. Linguistics 17: 79-106.
Dechert, H. 1983. ‘How a story is done in a second language’ in Fserch and Kasper (eds.) 1983a.
Dechert, H. 1984a. ‘Individual variation in language’ in Dechert et al. (eds.) 1984.
Dechert, H. 1984b. ‘Second language production: six hypotheses’ in Dechert et al. (eds.) 1984.
Dechert, H. (ed.). 1990. Current Trends in European Second Language Acquisition Research.
Clevedon, Avon: Multilingual Matters.
Dechert, H., D. Mohle, and M. Raupach (eds.). 1984. Second Language Productions. Tubingen:
Gunter Narr.
Dechert, H. and M. Raupach (eds.). 1989a. Transfer in Language Production. Norwood, N.J.:
Ablex.
Dechert, H. and M. Raupach (eds.). 1989b. Interlingual Processes. Tubingen: Gunter Narr.
de Jong, J. and D. Stevenson (eds.). 1990. Individualizing the Assessment of Language Liabil¬
ities. Clevedon, Avon: Multilingual Matters.
Derwing, T. 1989. ‘Information type and its relation to nonnative speaker comprehension’. Lan¬
guage Learning 39:157-72.
de Villiers, J. and P. de Villiers. 1973. ‘A cross-sectional study of the development of grammatical
morphemes in child speech ’.Journal of Psych olinguistic Research 1: 299-310.
Dickerson, L. 1974. ‘Internal and external patterning of phonological variability in the speech of
Japanese learners of English’. Unpublished PhD thesis, Urbana: University of Illinois.
Dickerson. L. 1975. ‘The learner’s interlanguage as a system of variable rules’. TESOL Quar¬
terly 9:401-7.
Dickinson, L. 1987. Self-instruction in Language Learning. Cambridge: Cambridge University
Press.
Diller, K. 1978. The Language Teaching Controversy. Rowley, Mass.: Newbury House.
Diller K. (ed.). 1981. Individual Differences and Universals in Language Learning Aptitude.
Rowley, Mass.: Newbury House.
Dittmar, N. 1981. ‘On the verbal organization of L2 tense marking in an elicited translation task
by Spanish immigrants in Germany’. Studies in Second Language Acquisition 3:136-64.
Dittmar, N. 1992. ‘Grammaticalization in second language acquisition: Introduction’. Studies
in Second Language Acquisition 14: 249-59.
Dittmar, N., H. Haberland, T. Skuttnab-Kangas, and U. Telman (eds.). 1978. Papers from the
First Scandinavian-German Symposium on the Language of Immigrant Workers and their
Children. Linguistgruppen, Roskilde Universiteits Center.
Dornyei, Z. 1990. ‘Conceptualising motivation in foreign language learning’. Language Learn¬
ing 40: 45-78.
Doughty, C. 1991. ‘Second language instruction does make a difference: evidence from an em¬
pirical study on SL relativization’. Studies in Second Language Acquisition 13: 431-69.
Doughty, C. and T. Pica. 1986. ‘ “Information gap” tasks: do they facilitate second language
acquisition?’ TESOL Quarterly 20: 305-25.
Dowd, J., J. Zuengler, and D. Berkowitz. 1990. ‘L2 social marking: research issues’. Applied
Linguistics 11: 16-29.
Downes, N. 1981. ‘Foreigner talk inside and outside the classroom’. Unpublished paper, Depart¬
ment of Linguistics, University of Pittsburgh. Cited in Chaudron 1988.
Duda, R. and P. Riley (eds.). 1990. Learning Styles. Nancy, France: University of Nancy.
Dudley-Evans, A. and T. Johns. 1981. ‘A team teaching approach to lecture comprehension for
overseas students’ in ELT Documents Special: The Teaching of Listening Comprehension.
London: The British Council.
 Bibliography 745

Duff, P. 1986. ‘Another look at mterlanguage talk: taking task to task’ in Day (ed.) 1986.
Dulay, H. and M. Burt. 1972. ‘Goofing, an indicator of children’s second language strategies’.
Language Learning 22: 234-52.
Dulay, H. and M. Burt. 1973. ‘Should we teach children syntax?’ Language Learning 23-
245-58.
Dulay, H. and M. Burt. 1974a. ‘You can’t learn without goofing’ in Richards (ed.) 1974.
Dulay, H. and M. Burt. 1974b. ‘Errors and strategies in child second language acquisition’.
TESOL Quarterly 8:129-36.
Dulay, H. and M. Burt. 1974c. ‘Natural sequences in child second language acquisition’. Lan¬
guage Learning 24: 37-53.
Dulay, H. and M. Burt. 1974d. ‘A new perspective on the creative construction processes in child
second language acquisition’. Language Learning 24: 253-78.
Dulay, H. and M. Burt. 1975. ‘Creative construction in second language learning and teaching’
in Burt and Dulay (eds.) 1975.
Dulay, H. and M. Burt. 1977. ‘Remarks on creativity in language acquisition’ in Burt et al. (eds.)
1977.
Dulay, H. and M. Burt. 1980. ‘On acquisition orders’ in Felix (ed.) 1980a.
Dulay, H., M. Burt, and S. Krashen. 1982. Language Two. New York: Oxford University Press.
Dunkel, H. 1948. Second Language Learning. Boston: Ginn.
Duskova, L. 1969. ‘On sources of errors in foreign language learning’. International Review of
Applied Linguistics 7: 11-36.

Early, M. 1985. ‘Input and interaction in content classrooms: foreigner talk and teacher talk in
classroom discourse’. Unpublished PhD thesis. University of California at Los Angeles.
Eckman, F. 1977. ‘Markedness and the contrastive analysis hypothesis’. Language Learning 27:
315-30.
Eckman, F. 1981. ‘On the naturalness of interlanguage phonological rules’. Language Learning
31: 195-216.
Eckman, F. 1984. ‘Universals, typologies and interlanguage’ in Rutherford (ed.) 1984a.
Eckman, F. 1985. ‘Some theoretical and pedagogical implications of the markedness differential
hypothesis’. Studies in Second Language Acquisition 7: 289-307.
Eckman, F. 1991. ‘The structural conformity hypothesis and the acquisition of consonant clus¬
ters in the interlanguage of ESL learners’. Studies in Second Language Acquisition 13:
23-41.
Eckman, F, L. Bell, and D. Nelson (eds.). 1984. Universals of Second Language Acquisition.
Rowley, Mass.: Newbury House.
Eckman, F., L. Bell, and D. Nelson. 1988. ‘On the generalization of relative clause instruction in
the acquisition of English as a second language’. Applied Linguistics 9:1-20.
Edelsky, C., B. Altwerger, F. Barkin, B. Flores, S. Hudleson, and K. Jilbert. 1983. ‘Semilingual-
ism and language deficit’. Applied Linguistics 5:113-27.
Edmondson, W. 1985. ‘Discourse worlds in the classroom and in foreign language’. Studies in
Second Language Acquisition 7: 159-68.
Edmondson, W. and J. House. 1991. ‘Do learners talk too much? The waffle phenomenon in
interlanguage pragmatics’ in Phillipson et al. (eds.) 1991.
Edwards, G. 1977. Second language retention in the Canadian Public Service. Ottowa: Public
Service Commission of Canada.
Edwards, J. 1984. ‘Irish: planning and preservation’. Journal of Multilingual and Multicultural
Development 5: 267-75.
Ehrlich, S. P. Avery, and C. Yorio. 1989. ‘Discourse structure and the negotiation of compre¬
hensible input’. Studies in Second Language Acquisition 11: 397-414.
Ehrman, M. 1990. ‘The role of personality type in adult language learning: an ongoing investiga¬
tion’ in Parry and Stansfield (eds.) 1990.
746 Bibliography

Ehrman, M. and R. Oxford. 1989. ‘Effects of sex differences, career choice, and psychological
type on adult language learning strategies’. Modern Language Journal 73:1-13.
Eisenstein, M. 1980. ‘Grammatical explanations in ESL: Teach the student, not the method’.
TESL Talk 11:3-13.
Eisenstein, M. 1982. ‘A study of social variation in adult second language acquisition’. Language
Learning 32: 367-92.
Eisenstein, M. 1983. ‘Native reactions to non-native speech: a review of empirical research .
Studies in Second Language Acquisition 5: 160-76.
Eisenstein, M. (ed.). 1989. The Dynamic Interlanguage: Empirical Studies in Second Language
Variation. New York: Plenum Press.
Eisenstein, M., N. Bailey, and C. Madden. 1982. ‘It takes two: contrasting tasks and contrasting
structures’. TESOL Quarterly 16: 381-93.
Eisenstein, M. and J. Bodman. 1986. ‘“I very appreciate”: expressions of gratitude by native and
non-native speakers of American English’. Applied Linguistics 7: 167-85.
Eisenstein, M. and G. Verdi. 1985. ‘The intelligibility of social dialects for working-class adult
learners of English’. Language Learning 35: 287-98.
Ekstrand, L. 1977. ‘Social and individual frame factors in L2 learning: comparative aspects’ in
Skuttnab-Kangas (ed.) 1977.
Ellis, G. and B. Sinclair. 1989. Learning to Learn English: a Course in Learner Training. Cam¬
bridge: Cambridge University Press.
Ellis, N. 1991. ‘Rules and instances in foreign language learning: interactions of explicit and im¬
plicit knowledge’. Unpublished paper, Bangor: University College of North Wales.
Ellis, R. 1980. ‘Classroom interaction and its relation to second language learning’. RELC
Journal 11: 29-48.
Ellis, R. 1982. ‘The origins of interlanguage’. Applied Linguistics 3: 207-23.
Ellis, R. 1984a. Classroom Second Language Development. Oxford: Pergamon.
Ellis, R. 1984b. ‘Formulaic speech in early classroom second language development’ in Hand-
scombe et al. (eds.) 1984.
Ellis, R. 1984c. ‘Can syntax be taught? A study of the effects of formal instruction on the acquisi¬
tion of Wh-questions by children’. Applied Linguistics 5:138-55.
Ellis, R. 1985a. Understanding Second Language Acquisition. Oxford: Oxford University Press.
Ellis, R. 1985b. ‘The L2=L1 Hypothesis: a reconsideration’. System 13: 9-24.
Ellis, R. 1985c. ‘Sources of variability in interlanguage’. Applied Linguistics 6: 118-31.
Ellis, R. 1985d. ‘Teacher-pupil interaction in second language development’ in Gass and Mad¬
den (eds.) 1985.
Ellis, R. (ed.). 1987a. Second Language Acquisition in Context. London: Prentice-Hall
International.
Ellis, R. 1987b. ‘Interlanguage variability in narrative discourse: style-shifting in the use of the
past tense’. Studies in Second Language Acquisition 9: 1-20.
Ellis, R. 1987c. ‘Contextual variability in second language acquisition and the relevancy of lan¬
guage teaching’ in Ellis (ed.) 1987a.
Ellis, R. 1988a. ‘The effects of linguistic environment on the second language acquisition of
grammatical rules’. Applied Linguistics 9: 257-74.
Ellis, R. 1988b. ‘The role of practice in classroom language learning’. AILA Review 5: 20-39.
Ellis, R. 1989a. ‘Classroom learning styles and their effect on second language acquisition: a
study of two learners’. System 17: 249-62.
Ellis, R. 1989b. ‘Are classroom and naturalistic acquisition the same? A study of the classroom
acquisition of German word order rules’. Studies in Second Language Acquisition 11:
305-28.
Ellis, R. 1989c. ‘Sources of intra-learner variability in language use and their relationship to
second language acquisition’ in Gass et al. (eds.) 1989b.
Ellis, R. 1990a. Instructed Second Language Acquisition. Oxford: Blackwell.
 Bibliography 747

Ellis, R. 1990b. ‘Individual learning styles in classroom second language development’ in de Jong
and Stevenson (eds.) 1990.
Ellis, R. 1990c. ‘A response to Gregg’. Applied Linguistics 11: 384-91.
Ellis, R. 1990d. ‘Grammaticality judgements and learner variability’ in Burmeister and Rounds
(eds.) 1990.
Ellis, R. 1991a. ‘The interaction hypothesis: a critical evaluation’ in Sadtono (ed.) 1991.
Ellis, R. 1991c. ‘Grammaticality judgments and second language acquisition’. Studies in Second
Language Acquisition 13: 161-86.
Ellis, R. 1991d. ‘Grammar teaching—practice or consciousness-raising’ in R. Ellis. Second Lan¬
guage Acquisition and Second Language Pedagogy. Clevedon, Avon: Multilingual Matters.
Ellis, R. 1992a. ‘Learning to communicate in the classroom’. Studies in Second Language Ac¬
quisition 14: 1-23.
Ellis, R. 1992b. ‘On the relationship between formal practice and second language acquisition’.
Die Neueren Sprachen 91: 131-47.
Ellis, R. 1993a. ‘Second language acquisition and the structural syllabus’. TESOL Quarterly 27:
91-113.
Ellis, R. 1993b. ‘Interpretation-based grammar teaching’. System 21: 69-78.
Ellis, R. 1993c. ‘Second language acquisition research and teacher development: the case of
teachers’ questions’. Paper given at Second International Conference on Teacher Education
in Second Language Teaching, City Polytechnic of Hong Kong.
Ellis, R. and M. Rathbone. 1987. The Acquisition of German in a Classroom Context. Mimeo¬
graph. London: Ealing College of Higher Education.
Ellis, R. and C. Roberts. 1987. ‘Two approaches for investigating second language acquisition in
context’ in Ellis (ed.) 1987a.
Ellis, R. and G. Wells. 1980. ‘Enabling factors in adult-child discourse’. First Language 1:46-82.
Ely, C. 1986a. An analysis of discomfort, risktaking, sociability, and motivation in the L2 class¬
room’. Language Learning 36: 1-25.
Ely, C. 1986b. ‘Language learning motivation: a descriptive and causal analysis’. Modern Lan¬
guage Journal 70: 28-35.
Enright, D. 1984. ‘The organization of interaction in elementary classrooms’ in Handscombe et
al. (eds.) 1984.
Epstein, N. 1977. Language, Ethnicity and the Schools. Washington D.C.: Institute for Educa¬
tional Leadership.
Erickson, F. and G. Mohatt. 1982. ‘Cultural organization of participation structures in two
classrooms of Indian students’ in Spindler (ed.) 1982.
Erickson, F. and J. Schultz. 1982. The Counselor as Gatekeeper: Social Interaction in Interviews.
New York: Academic Press.
Ervin-Tripp, S. 1974. ‘Is second language learning like the first?’ TESOL Quarterly 8:111-27.
Ervin-Tripp, S., A. Strage, M. Lampert, and N. Bell. 1987. ‘Understanding requests’. Linguistics
25:107-43.
Eubank, L. 1987a. ‘Parameters in L2 learning: Flynn revisited’. Second Language Research 5:
43-73.
Eubank, L. 1987b. ‘The acquisition of German negation by formal language learners’ in Van-
Patten et al. (eds.) 1987.
Eubank, L. 1990. ‘Linguistic theory and the acquisition of German negation’ in VanPatten and
Lee (eds.) 1990.
Eubank, L. 1991a. ‘Introduction’ in L. Eubank (ed.) 1991b.
Eubank, L. (ed.). 1991b. Point Counterpoint: Universal Grammar in the Second Language.
Amsterdam: John Benjamins.
Eysenck, H. 1970. The Structure of Human Personality. London: Routledge, Kegan, and Paul.
Eysenck, S. and J. Chan. 1982. ‘A comparative study of personality in adults and children: Hong
Kong vs. England’. Personality and Individual Differences 3: 153-60.
748 Bibliography

Fserch, C. 1980. ‘Describing interlanguage through interaction: problems of systematicity and

permeability’. Working Papers on Bilingualism 19: 59-78.
Fserch, C. 1985. ‘Meta talk in FL classroom discourse’. Studies in Second Language Acquisition
7:184-99.
Fserch, C. and G. Kasper. 1980. ‘Processes and strategies in foreign language learning and
communication’. Interlanguage Studies Bulletin 5: 47-118.
Fxrch, C. and G. Kasper (eds.). 1983a. Strategies in Interlanguage Communication. London:
Longman.
Fserch, C. and G. Kasper. 1983b. ‘Plans and strategies in foreign language communication’ C.
Fserch and G. Kasper (eds.) 1983a.
Fserch, C. and G. Kasper. 1985. ‘Procedural knowledge as a component of foreign language
learners’ communicative competence’ in Bolte and Flerrlitz (eds.) 1985.
Fserch C. and G. Kasper. 1986a. ‘The role of comprehension in second language acquisition’.
Applied Linguistics 7:257-74.
Fserch, C. and G. Kasper. 1986b. ‘Cognitive dimensions of language transfer’ in Kellerman and
Sharwood Smith (eds.) 1986.
Fserch, C. and G. Kasper (eds.). 1987. Introspection in Second Language Research. Clevedon,
Avon: Multilingual Matters.
Fserch, C. and G. Kasper. 1989. ‘Internal and external modification in interlanguage request
realization’ in Blum-Kulka et al. (eds.) 1989a.
Fanselow, J. 1977a. ‘Beyond “Rashomon”—conceptualizing and describing the teaching act’.
TESOL Quarterly 10: 17-39.
Fanselow, J. 1977b. ‘The treatment of error in oral work’. Foreign Language Annals 10:583-93.
Fanselow, J. and R. Crymes (eds.). 1976. On TESOL ’76. Washington D.C.: TESOL.
Fasold, R. 1984. ‘Variation theory and language learning’ in Trudgill (ed.) 1984.
Fasold, R. and R. Shuy (eds.). 1975. Analyzing Variation in Language. Washington D.C.:
Georgetown University Press.
Fathman, A. 1975. ‘Language background, age, and the order of acquisition of English
structures’ in Burt and Dulay (eds.) 1975.
Fathman, A. 1976. ‘Variables affecting the successful learning of English as a second language’.
TESOL Quarterly 10: 433-41.
Fathman, A. 1978. ‘ESL and EFI, learning:,similar or dissimilar?’ in Blatchford and Schachter
(eds.) 1978.
Felix, S. 1978. ‘Some differences between first and second language acquisition’ in Waterson and
Snow (eds.) 1978.
Felix, S. (ed.). 1980a. Second Language Development: Trends and Issues. Tubingen: Gunter
Narr.
Felix, S. 1980b. ‘Interference, interlanguage and related issues’ in Felix (ed.) 1980a.
Felix, S. 1981. ‘The effect of formal instruction on second language acquisition’. Language
Learning 31: 87-112.
Felix, S. 1984. ‘Maturational aspects of Universal Grammar’ in Davies et al. (eds.) 1984.
Felix, S. 1985. ‘More evidence on competing cognitive systems’. Second Language Research 1:
47-72.
Felix, S. and A. Flahn. 1985. ‘Natural processes in classroom second-language learning’. Applied
Linguistics 6: 223-38.
Felix, S. and W. Weigh 1991. ‘Universal grammar in the classroom: the effects of formal instruc¬
tion on second language acquisition’. Second Language Research 7: 162-80.
Ferguson, C. 1971. ‘Absence of copula and the notion of simplicity: a study of normal speech,
baby talk, foreigner talk and pidgins’ in Hymes (ed.) 1971b.
Ferguson, C. 1975. ‘Towards a characterization of English foreigner talk’. Anthropological Lin¬
guistics 17: 1-14.
Ferguson, C. 1977. ‘Baby talk as a simplified register’ in Snow and Ferguson (eds.) 1977.
Ferguson, C. and C. Debose. 1977. ‘Simplified registers, broken languages and pidginization’ in
Valdman (ed.) 1977.
 Bibliography 749

Ferguson, C. and D. Slobin (eds.). 1973. Studies of Child Language Development. New York:
Appleton-Century-Crofts.
Ferrier, L. 1978. ‘Some observations of error in context’ in Waterson and Snow (eds.) 1978.
Fiksdal, S. 1989. ‘Framing uncomfortable moments in crosscultural gatekeeping interviews’ in
Gass et al. (eds.) 1989a.
Fillmore, C. 1968. ‘The case for case’ in Bach and Harms (eds.) 1968.
Filmore, C., D. Kempler and W. Wang (eds.). 1979. Individual Differences in Language Ability
and Language Behavior. New York: Academic Press.
Fine, J. (ed.). 1988. Second Language Discourse: a Textbook of Current Research. Norwood,
N.J.: Ablex.
Finer, D. 1991. ‘Binding parameters in second language acquisition’ in Eubank (ed.) 1991b.
Finocchiaro, M. 1981. ‘Motivation: its crucial role in language learning’ in Hines and Ruther¬
ford (eds.) 1981.
Fisher, J., M. Clarke, and J. Schachter (eds.) 1980. On TESOL ’80. Washington D.C.: TESOL.
Fisherman, J., C. Ferguson, and J. Das (eds.). 1968. Language Problems of Developing Nations.
New York: Wiley and Sons.
Fisiak, J. 1981a. ‘Some introductory notes concerning contrastive linguistics’ in Fisiak (ed.).
Fisiak, J. (ed.). 1981b. Contrastive Linguistics and the Language Teacher. Oxford: Pergamon.
Fischer, J. 1958. ‘Social influences in the choice of a linguistic variant’. Word 14: 47-56.
Fishman, J., R. Cooper, and A. Conrad. 1977. The Spread of English. Rowley: Mass.: Newbury
House.
Fishman, J., R. Cooper, and R. Ma (eds.). 1968. Bilingualism in the Barrio. New York: Yeshiva
University.
Fitzpatrick, F. 1987. The Open Door: The Bradford Bilingual Project. Clevedon, Avon: Multi¬
lingual Matters.
Flanders, N. 1970. Analyzing Teaching Behavior. Reading, Mass.: Addison-Wesley.
Flanigan, B. 1991. ‘Peer tutoring and second language acquisition in the elementary school’.
Applied Linguistics 12: 141-58.
Flavell,J. and M. Markman (eds.). 1983. Carmichael’s Manual of Child Psychology: Vo/3. New
York: Wiley and Sons.
Flege, J. 1987. ‘A critical period for learning to pronounce foreign languages’. Applied Linguist¬
ics 8: 162-77.
Fletcher, P. and M. Garman (eds.). 1986. Language Acquisition (2nd ed.). Cambridge: Cam¬
bridge University Press.
Flick, W. 1979. ‘A multiple component approach to research in second language acquisition’ in
Andersen (ed.) 1979a.
Flick, W. 1980. ‘Error types in adult English as a second language’ in Ketterman and St. Clair
(eds.) 1980.
Flores d’Arcais, G. and W. Levelt (eds.). 1970. Advances in Psycholinguistics. Amsterdam:
North-Holland Publishing.
Flowerdew, J. 1992. ‘Definitions in science lectures’. Applied Linguistics 13: 202-21.
Flynn, S. 1984. ‘A universal in L2 acquisition based on a PBD typology’ in Eckman et al. (eds.)
1984.
Flynn, S. 1987. A parameter-setting model ofL2 Acquisition. Dordrecht: Reidel.
Flynn, S. and B. Lust. 1990. ‘In defense of parameter-setting in L2 acquisition: a reply to Bley-
Vroman and Chaudron ’90’. Language Learning 40: 419-49.
Flynn, S. and S. Manuel. 1991. ‘Age-dependent effects in language acquisition: an evaluation of
the “critical period” hypothesis’ in Eubank (ed.) 1991b.
Flynn, S. and W. O’Neill, (eds.). 1988. Linguistic Theory in Second Language Acquisition.
Dordrecht: Kluwer.
Forrest-Pressley, D., G. Mackinnon and T. Waller (eds.). 1985. Metacognition, Cognition, and
Human Performance, Vol 1. New York: Academic Press.
750 Bibliography

Foster, S. 1990. The Communicative Competence of Young Children: a Modular Approach.

London: Longman.
Fotos, S. 1993. ‘Consciousness-raising and noticing through focus on form: grammar task per¬
formance versus formal instruction’. Applied Linguistics 14: 4.
Fotos, S. and R. Ellis. 1991. ‘Communicating about grammar: a task-based approach’. TESOL
Quarterly 25: 605-28.
Fraser, B. 1981. ‘On apologizing’ in Coulmas (ed.) 1981.
Fraser, B. 1983. ‘The domain of pragmatics’ in Richards and Schmidt (eds.) 1983.
Fraser, B., E. Rintell, and J. Walters. 1980. ‘An approach to conducting research on the acquisi¬
tion of pragmatic competence in a second language’ in Larsen-Freeman (ed.) 1980.
Fraser, C. and K. Scherer (eds.). 1982. Social Psychological Dimensions of Language Behaviour.
Cambridge: Cambridge University Press.
Freed, B. 1980. ‘Talking to foreigners vs. talking to children: similarities and differences’ in Scar-
cella and Krashen (eds.) 1980.
Freed, B. 1981. ‘Foreigner talk, baby talk, native talk’. International Journal of the Sociology of
Language 28:19-39.
Freed, B. 1990. ‘Language learning in a study abroad context: the effects of interactive and non¬
interactive out-of-class contact on grammatical achievement and oral proficiency’ in Pro¬
ceedings of the Georgetown University Round Table on Languages and Linguistics 1990.
Washington D.C.: Georgetown University.
French, F. 1949. Common Errors in English. London: Oxford University Press.
French, P. and M. McLure (eds.). 1981. Adult-Child Conversation. New York: St Martin’s
Press.
Freudenstein, R. 1977. ‘Interaction in the foreign language classroom’ in Burt et al. (eds.) 1977.
Fromkin, V. 1971. ‘The non-anomalous nature of anomalous utterances’. Language 47: 27-52.
Fry,J. 1988. ‘Diary studies in classroom SLA: problems and prospects’. JALTJournal 9:158-67.
Fujimoto, D., J. Lubin, Y. Sasaki, and M. Long. 1986. ‘The effect of linguistic and conversational
adjustments on the comprehensibility of spoken second language discourse’. Unpublished
manuscript, Department of ESL, University of Hawaii at Manoa. Cited in Larsen-Freeman
and Long 1991.
Furrow, D., K. Nelson, and H. Benedict. 1979. ‘Mothers’ speech to children and syntactic devel¬
opment: some simple relationships’. Journal of Child Language 6: 423-42.

Gaies, S. 1977. ‘The nature of linguistic input in formal second language learning: linguistic and
communicative strategies’ in Brown et al. (eds.) 1977.
Gaies, S. 1982. ‘Native speaker-nonnative speaker interaction among academic peers’. Studies in
Second Language Acquisition 5: 74-82.
Gaies, S. 1983a. ‘The investigation of language classroom processes’. TESOL Quarterly 17:
205-18.
Gaies, S. 1983b. ‘Learner feedback: an exploratory study of its role in the second language class¬
room’ in Seliger and Long (eds.) 1983.
Gardner, R. 1979. ‘Social psychological aspects of second language acquisition’ in Giles and St.
Clair (eds.) 1979.
Gardner, R. 1980. ‘On the validity of affective variables in second language acquisition: concep¬
tual, contextual, and statistical considerations’. Language Learning 30: 255-70.
Gardner, R. 1983. ‘Learning another language: a true social psychological experiment’. Journal
of Language and Social Psychology 2: 219-40.
Gardner, R. 1985. Social Psychology and Second Language Learning: The Role of Attitude and
Motivation. London: Edward Arnold.
Gardner, R. 1988. ‘The socio-educational model of second language learning: assumptions, find¬
ings and issues’. Language Learning 38: 101-26.
 Bibliography 751

Gardner, R. 1991. ‘Second-language learning in adults: correlates of proficiency’. Applied Lan¬

guage Learning 2: 1-28.
Gardner, R. and R. Clement. 1990. ‘Social psychological perspectives on second language ac¬
quisition’ in Giles and Robinson (eds.) 1990.
Gardner, R., R. Ginsberg, and P. Smythe. 1976. ‘Attitudes and motivation in second language
learning: course related changes’. The Canadian Modern Language Review 32: 243-66.
Gardner, R., R. Lalonde, and J. MacPherson. 1985. ‘Social factors in second language attrition’.
Language Learning 35: 519-40.
Gardner, R., R. Lalonde, and R. Pierson. 1983. ‘The socio-educational model of second lan¬
guage acquisition: an investigation using LISREL causal modeling’. Journal of Language
and Social Psychology 2: 1-15.
Gardner, R. and W. Lambert. 1959. ‘Motivational variables in second language acquisition’.
Canadian Journal of Psychology 13: 266-72.
Gardner, R. and W. Lambert. 1972. Attitudes and Motivation in Second Language Learning.
Rowley, Mass.: Newbury House.
Gardner, R. and P. MacIntyre. 1991. ‘An instrumental motivation in language study: who says it
isn’t effective?’ Studies in Second Language Acquisition 13: 57-72.
Gardner, R. and P. MacIntyre. 1992. ‘A student’s contributions to second language learning.
Part 1: Cognitive variables’. Language Teaching 25: 211-20.
Gardner, R., R. Moorcroft, and P. MacIntyre. 1987. ‘The role of anxiety in second language per¬
formance of language dropouts’. Research Bulletin No 657. London, Ontario: The Univer¬
sity of Western Ontario.
Gardner, R. and P. Smythe. 1975. ‘Second language acquisition: a social psychological ap¬
proach’. Research Bulletin No 332. Department of Psychology, University of Western
Ontario. Cited in Gardner 1985.
Gardner, R., P. Smythe, and G. Brunet. 1977. ‘Intensive second language study: effects on atti¬
tudes, motivation and French achievement’. Language Learning 27: 243-62.
Gardner, R., P. Smythe, and R. Clement. 1979. ‘Intensive second language study in a bicultural
milieu: an investigation of attitudes, motivation, and language proficiency’. Language
Learning 29: 305-20.
Gardner, R., P. Smythe, R., Clement, and L. Gliksman. 1976. ‘Second language lcarning: a
social-psychological perspective’. Canadian Modern Language Review 32:198-213.
Garnica, O. 1977. ‘Some prosodic and paralinguistic features of speech to young children’ in
Snow and Ferguson (eds.) 1977.
Gaskill, W. 1980. ‘Correction in native speaker - nonnative speaker conversation’ in Larsen-
Freeman (ed.) 1980.
Gass, S. 1979. ‘Language transfer and universal grammatical relations’. Language Learning 29:
327-44.
Gass, S. 1980. ‘An investigation of syntactic transfer in adult second language learners’ in Scar-
cella and Krashen (eds.) 1980.
Gass, S. 1982. ‘From theory to practice’ in Hines and Rutherford (eds.) 1981.
Gass, S. 1983. ‘Language transfer and universal grammatical relations’ in Gass and Selinker (ed.)
1983.
Gass, S. 1984. ‘A review of interlanguage syntax: language transfer and language universals’.
Language Learning 34: 115-32.
Gass, S. 1987. ‘The resolution of conflicts among competing systems: a bidirectional perspect¬
ive’. Applied Psycholinguistics 8: 329-50.
Gass, S. 1988. ‘Integrating research areas: a framework for second language studies’. Applied
Linguistics 9: 198-217.
Gass, S. 1989. ‘Language universals and second language acquisitfonATawg«a^ Learning 39:
497-534.
752 Bibliography

Gass, S. 1990. ‘Second and foreign language learning: same, different or none of the above?’ in
VanPatten and Lee (eds.) 1990.
Gass, S. 1991. ‘Grammar instruction, selective attention, and learning’ in Phillipson et al. (eds.)
1991.
Gass, S. and J. Ard. 1980. ‘L2 data: their relevance for language universals’. TESOL Quarterly
16:443-52.
Gass, S. and J. Ard. 1984. ‘Second language acquisition and the ontology of language universals’
in Rutherford (ed.) 1984a.
Gass, S. and G. Crookes (eds.). (forthcoming). Task-based Learning in a Second Language.
Clevedon, Avon: Multilingual Matters.
Gass, S. and U. Lakshmanan. 1991. ‘Accounting for interlanguage subject pronouns’. Second
Language Research 7: 181-203.
Gass, S. and C. Madden (eds.). 1985. Input in Second Language Acquisition. Rowley, Mass.:
Newbury House.
Gass, S., C. Madden, D. Preston, and L. Selinker (eds.). 1989a. Variation in Second Language
Acquisition Volume I: Sociolinguistic Issues. Clevedon, Avon: Multilingual Matters.
Gass, S., C. Madden, D. Preston, and L. Selinker (eds.). 1989b. Variation in Second Language
Acquisition Volume II: Psycholinguistic Issues. Clevedon, Avon: Multilingual Matters.
Gass, S. and J. Schachter (eds.). 1989. Linguistic Perspectives on Second Language Acquisition.
Cambridge: Cambridge University Press.
Gass, S. and L. Selinker (eds.). 1983. Language Transfer in Language Learning. Rowley, Mass.:
Newbury House.
Gass, S. and E. Varonis. 1985a. ‘Task variation and nonnative/nonnative negotiation of mean¬
ing’ in Gass and Madden (eds.) 1985.
Gass, S. and E. Varonis. 1985b. ‘Variation in native speaker speech modification to non-native
speakers’. Studies in Second Language Acquisition 7: 37-57.
Gass, S. and E. Varonis. 1986. ‘Sex differences in NNS/NNS interactions’ in Day (ed.) 1986.
Gass, S. and E. Varonis. 1991. ‘Miscommunication in nonnative speaker discourse’ in Coupland
et al. (eds.) 1991.
Gasser, M. 1990. ‘Connectionism and universals of second language acquisition’. Studies in
Second Language Acquisition 12: 179-99.
Gatbonton, E. 1978. ‘Patterned phonetic variability in second language speech: a gradual dif¬
fusion model’. Canadian Modern Language Review 34: 335-47.
Genesee, F. 1976. ‘The role of intelligence in second language learning’. Language Learning 26:
267-80.
Genesee, F. 1978. ‘Individual differences in second language learning’. The Canadian Modern
Language Review 34: 490-504.
Genesee, F. 1984. ‘French immersion programs’ in Shapson and D’Oyley (eds.) 1984.
Genesee, F. 1987. Learning Through Two Languages: Studies of Immersion and Bilingual
Education. Cambridge, Mass.: Newbury House.
George, H. 1972. Common Errors in Language Learning: Insights from English. Rowley, Mass.:
Newbury House.
Gerbault, J. 1978. ‘The acquisition of English by a five-year-old French speaker’. MA thesis,
University of California at Los Angeles.
Gibbons, J. 1985. ‘The silent period: an examination’. Language Learning 35: 255-67.
Gick, M. and K. Holyoak. 1983. ‘Schema induction and analogical transfer’. Cognitive Psychol¬
ogy 15: 1-38.
Gieve, S. 1991. ‘Goals and preferred language styles of Japanese English majors’ in Proceedings
of the Conference on Second Language Research in Japan. International University of
Japan.
Giles, H. 1971. ‘Our reactions to accent’. New Society, 14th October.
 Bibliography 753

Giles, H. (ed.). 1977. Language, Ethnicity and Intergroup Relations. New York: Academic
Press.
Giles, H. and J. Byrne. 1982. ‘An intergroup approach to second language acquisition’. Journal
of Multicultural and Multilingual Development 3: 17-40.
Giles, H. and P. Johnson. 1981. ‘The role of language in ethnic group relations’ in Turner and
Giles (eds.) 1981.
Giles, H. and W. Robinson (eds.). 1990. Handbook of Language and Social Psychology. Chi¬
chester: John Wiley and Sons.
Giles, H., W. Robinson, and P. Smith (eds.). 1980. Language: Social Psychological Perspectives.
Oxford: Pergamon Press.
Giles, H. and E. Ryan. 1982. ‘Prolegomena for developing a social psychological theory of lan¬
guage attitudes’ in Ryan and Giles (eds.) 1982.
Giles, H. and R. St. Clair (eds.). 1979. Language and Social Psychology. Oxford: Blackwell.
Gillette, B. 1987. ‘Two successful language learners: an introspective report’ in Faerch and
Kasper (eds.) 1987.
Gingras, R. (ed.). 1978. Second Language Acquisition and Foreign Language Teaching. Arling¬
ton, VA.: Center for Applied Linguistics.
Givon, T. 1979a. On Understanding Grammar. New York: Academic Press.
Givon, T. 1979b. Syntax and Semantics, Vol 12: Discourse and Semantics. New York: Academic
Press.
Givon, T. 1984. ‘Universals of discourse structure and second language acquisition’ in Ruther¬
ford (ed.) 1984a.
Givon, T. 1995. Functionalism and Grammar. Amsterdam: J.Benjamins.
Glahn, E. and A. Flolmen (eds.). 1985. Learner Discourse. Anglica et Americana 22. Copen¬
hagen: University of Copenhagen.
Glaser, R. (ed.). 1965. Training, Research, and Education. New York: Wiley and Sons.
Gleason, J. and S. Weintraub. 1978. ‘Input language and the acquisition of communicative com¬
petence’ in Nelson (ed.) 1978.
Gleitman, L., E. Newport, and H. Gleitman. 1984. ‘The current status of the motherese hypo¬
thesis’. Journal of Child Language 11: 43-79.
Gliksman, L. 1976. ‘Second language acquisition: the effects of student attitudes on classroom
behavior’. Unpublished MA thesis, University of Western Ontario.
Gliksman, L., R. Gardner, and P. Smythe. 1982. ‘The role of integrative motivation on students’
participation in the French classroom’. Canadian Modern Language Review 38: 625-47.
Godfrey, D. 1980. ‘A discourse analysis of tense’ in Larsen-Freeman (ed.) 1980.
Goldin-Meadow, S. 1982. ‘The resilience of recursion: a study of a communication system de¬
veloped without a conventional language model’ in Wanner and Gleitman (eds.) 1982.
Goldman-Eisler, F. 1968. Psycholinguistics: Experiments in Spontaneous Speech. New York:
Academic.
Goldstein, L. 1987. ‘Standard English: the only target for nonnative speakers of English?’
TESOL Quarterly 21: 417-36.
Goodluck, H. 1986. ‘Language acquisition and linguistic theory’ in Fletcher and Garman (eds.)
1986.
Goody, E. (ed.). 1978. Questions and Politeness. Cambridge: Cambridge University Press.
Grauberg, W. 1971. ‘An error analysis in the German of first-year university students’ in Perren
and Trim (eds.) 1971.
Green, P. 1975. ‘Aptitude testing: an ongoing experiment’. Audio-Visual Language journal 12:
205-10.
Green, P. and K. Hecht. 1992. ‘Implicit and explicit grammar: an empirical study’. Applied
Linguistics 13: 168-84.
Greenberg, H. 1966. Universals of Language (2nd Ed.). Cambridge, Mass.: MIT Press.
Greenfield, P. and C. Dent. 1980. ‘A developmental study of the communication of meaning: the
role of uncertainty of information’ in Nelson (ed.) 1980.
754 Bibliography

Gregg, K. 1984. ‘Krashen’s Monitor and Occam’s Razor’. Applied Linguistics 5: 79-100.
Gregg, K. 1989. ‘Second language acquisition theory: the case for a generative perspective’ in
Gass and Schachter (eds.) 1989.
Gregg, K. 1990. ‘The variable competence model of second language acquisition and why it
isn’t’. Applied Linguistics 11: 364-83.
Gregg, K. 1993. ‘Taking explanation seriously; or, let a couple of flowers bloom’. Applied Lin¬
guistics 14:276-94.
Gremmo, M., H. Holec, and P. Riley. 1977. ‘Interactional structure: the role of role’. Melanges
Pedagogiques, University of Nancy: CRAPEL.
Gremmo, M., H. Holec, and P. Riley. 1978. ‘Taking the initiative: some pedagogical applica¬
tions of discourse analysis’. Melanges Pedagogiques, University of Nancy: CRAPEL.
Griffiths, R. 1990. ‘Speech rate and NNS comprehension: a preliminary study in time-benefit
analysis’. Language Learning 40: 311-36.
Griffiths, R. 1991a. ‘Pausological research in an L2 context: a rationale and review of selected
studies’. Applied Linguistics 12: 345-64.
Griffiths, R. 1991b. ‘Personality and second-language learning: theory, research and practice’ in
Sadtano (ed.) 1991.
Griffiths, R. and R. Sheen. 1992. ‘Disembedded figures in the landscape: a reappraisal of L2
research on field dependence/independence’. Applied Linguistics 13:133-48.
Gudykunst, W. and S. Ting-Toomey. 1990. ‘Ethnic identity, language and communication
breakdowns’ in Giles and Robinson (eds.) 1990.
Guiora, A., W. Acton, R. Erard, and F. Strickland. 1980. ‘The effects of benzodiazepine (valium)
on permeability of language ego boundaries’. Language Learning 30: 351-63.
Guiora, A., B. Beit-Hallahmi, R. Brannon, C. Dull, and T. Scovel. 1972. ‘The effects of experi¬
mentally induced changes in ego states on pronunciation ability in a second language: an
exploratory study’. Comprehensive Psychiatry 13: 421-8.
Guiora, A., H. Lane, and L. Bosworth. 1967. ‘An explanation of some personality variables in
authentic pronunciation in a second language’ in Lane and Zale (eds.) 1967.
Gumperz, J. 1982a. Discourse strategies. Cambridge: Cambridge University Press.
Gumperz, J. (ed.). 1982b. Language and Social Identity. Cambridge: Cambridge University
Press.
Gundel, J. and E. Tarone. 1983. ‘Language transfer and the acquisition of pronominal ana¬
phora’ in Gass and Selinker (eds.) 1983.
Guy, G. 1988. ‘Language and social class’ in Neymeyer (ed.) 1988.

Haastrup, K. 1987. ‘Using thinking aloud and retrospection to uncover learner’s lexical in-
ferencing procedures’ in Fserch and Kasper (eds.) 1987.
Hakansson, G. 1986. ‘Quantitative studies of teacher talk’ in Kasper (ed.) 1986.
Hakansson, G. and I. Lindberg. 1988. ‘What’s the question? Investigating second language class¬
rooms’ in Kasper (ed.) 1988.
Hakuta, K. 1974. ‘A preliminary report on the development of grammatical morphemes in a
Japanese girl learning English as a second language’. Working Papers on Bilingualism 3:
18-43.
Hakuta, K. 1976. ‘A case study of a Japanese child learning English as a second language’.
Language Learning 26: 321-51.
Hakuta, K. and E. Cancino. 1977. ‘Trends in second language acquisition research’. Harvard
Educational Review 47: 294-316.
Hale, T. and E. Budar. 1970. ‘Are TESOL classes the only answer?’ Modern Language Journal
54:487-92.
Hall, B. and W. Gudykunst. 1986. ‘The intergroup theory of second language ability’. Journal of
Language and Social Psychology 5: 291-302.
Halliday, M. 1973. Explorations in the Functions of Language. London: Edward Arnold.
 Bibliography 755

Halliday, M. 1978. Language as a Social Semiotic. London: Edward Arnold.

Hamayan, E. and R. Tucker. 1980. ‘Language input in the bilingual classroom and its relations
to second language achievement’. TESOL Quarterly 14: 453-68.
Hammarberg, B. 1973. ‘The insufficiency of error analysis’ in Svartvik (ed.) 1973a.
Hammarberg, B. 1979. ‘On intralingual, interlingual and developmental solutions in inter¬
language’ in Hyltenstam and Linnarud (eds.) 1979.
Hammerley, H. 1975. ‘The deduction induction controversy’. Modern Language Journal 59-
15-18.
Hammerley, H. 1987. ‘The immersion approach: litmus test of second language acquisition
through classroom communication’. Modern Language Journal 71: 395-401.
Hammerley, H. 1989. French Immersion: Myths and Reality. Calgary, Alberta: Detselig
Enterprises.
Hammond, R. 1988. ‘Accuracy versus communicative competency: the acquisition of grammar
in the second language classroom’. Hispania 71: 408-17.
Hanania, E. and H. Gradman. 1977. ‘Acquisition of English structures: a case study of an adult
native speaker of Arabic in an English-speaking environment’. Language Learning 27:
75-91.
Handscombe, J., R. Orem, and B. Taylor (eds.). 1984. On TESOL ’83: The Question of Con¬
trol. Washington D.C.: TESOL.
Hansen, J. and C. Stansfield. 1981. ‘The relationship of field dependent-independent cognitive
styles to foreign language achievement’. Language Learning 31: 349-67.
Hansen, L. 1984. ‘Field dependence-independence and language testing: evidence from six
Pacific island cultures’. TESOL Quarterly 18: 311-24.
Hansen-Strain, L. and J. Strain. 1989. ‘Variation in the relative clause of Japanese learners’.
J ALT Journal 11:211-37.
Harder, P. 1980. ‘Discourse as self-expression—on the reduced personality of the second lan¬
guage learner’. Applied Linguistics 1: 262-70.
Harkness, S. 1977. ‘Aspects of social environment and first language acquisition in rural Africa’
in Snow and Ferguson (eds.) 1977.
Harley, B. 1986. Age in Second Language Acquisition. Clevedon, Avon: Multilingual Matters.
Harley, B. 1988. ‘Accounting for patterns of lexical development in a classroom L2 environ¬
ment’. Paper prepared for proceedings of Explaining Interlanguage Development Confer¬
ence held in Melbourne, August 1987.
Harley, B. 1989. ‘Functional grammar in French immersion: a classroom experiment’. Applied
Linguistics 19: 331-59.
Harley, B., P. Allen, J. Cummins, and M. Swain (eds.). 1987. The Development of Bilingual Pro¬
ficiency: Final Report, Vol III: Social Context and Age. Toronto: Modern Languages Cen¬
ter, IOSE.
Harley, B., P. Allen, J. Cummins, and M. Swain (eds.). 1990. The Development of Second Lan¬
guage Proficiency. Cambridge: Cambridge University Press.
Harley, B. and M. Swain. 1978. ‘An analysis of the verb system by young learners of French’.
Interlanguage Studies Bulletin 3: 35-79.
Harrington, M. 1987. ‘Processing transfer: language-specific processing strategies as a source of
interlanguage variation’. Applied Psycholinguistics 8: 351-77.
Hartnett, D. 1985. ‘Cognitive style and second language learning’ in Celce-Murcia (ed.) 1985.
Hatch, E. 1974. ‘Second language learning—universals?’ Working Papers on Bilingualism 3:
1-17.
Hatch, E. (ed.). 1978a. Second Language Acquisition. Rowley, Mass.: Newbury House.
Hatch, E. 1978b. ‘Discourse analysis and second language acquisition’ in Hatch (ed.) 1978a.
Hatch, E. 1978c. ‘Discourse analysis, speech acts and second language acquisition’ in Ritchie
(ed.) 1978a.
Hatch, E. 1978d. ‘Acquisition of syntax in a second language’ in Richards (ed.) 1978.
756 Bibliography

Hatch, E. 1978e. ‘Apply with caution’. Studies in Second Language Acquisition 2: 123-43.
Hatch, E. 1980. ‘Second language acquisition—avoiding the question’ in Felix (ed.) 1980a.
Hatch, E. 1983a. Psycholinguistics: A Second Language Perspective. Rowley, Mass.: Newbury
House.
Hatch, E. 1983b. ‘Simplified input and second language acquisition’ in Andersen (ed.) 1983c.
Hatch, E. 1992. Discourse and Language Education. Cambridge: Cambridge University Press.
Hatch, E. and H. Farhady. 1982. Research Design and Statistics for Applied Linguistics. Row-
ley, Mass.: Newbury House.
Hatch, E. and A. Lazaraton. 1991. The Research Manual: Design and Statistics for Applied Lin¬
guistics. New York: Newbury House/Harper Collins.
Hatch, E. and M. Long. 1980. ‘Discourse analysis, what’s that?’ in Larsen-Freeman (ed.) 1980.
Hatch, E., S. Peck, and J. Wagner-Gough. 1979. ‘A look at process in child second language ac¬
quisition’ in Ochs and Schieffelin (eds.) 1979.
Hatch, E., R. Shapira, and J. Wagner-Gough. 1978. ‘Foreigner talk discourse’. ITL Review of
Applied Linguistics 39/40: 39-60.
Hatch, E., Y. Shirai, and C. Fantuzzi. 1990. ‘The need for an integrated theory: connecting mod¬
ules’. TESOL Quarterly 24: 697-716.
Hatch, E. and J. Wagner-Gough. 1976. ‘Explaining sequence and variation in second language
acquisition’. Language Learning, Special Issue 4: 39-47.
Haugen, E. 1956. Bilingualism in the Americas. The American Dialect Society.
Hauptman, P. 1970. ‘An experimental comparison of a structural approach and a situational
approach to foreign language teaching’. Unpublished PhD thesis. Ann Arbor: University
of Michigan.
Hawkins, B. 1985. ‘Is the appropriate response always so appropriate?’ in Gass and Madden
(eds.) 1985.
Hawkins, J. 1983. Word Order Universals. New York: Academic Press.
Hawkins, R. 1987. ‘Markedness and the acquisition of the English dative alternation by L2
learners’. Second Language Research 3: 20-55.
Hawkins, R. 1988. ‘Comparing the effect of aural and written presentation of foreign language
material on learners’ memory for that material’. Unpublished paper, University of Sheffield.
Hawkins, R. 1989. ‘Do second language learners acquire restrictive relative clauses on the basis
of relational or configurational information? The acquisition of French subject, direct ob¬
ject and genitive restrictive relative clauses by second language learners’. Second Language
Research 5: 158-88.
Hayes,J. (ed.). 1970. Cognition and the Development of Language. New York: Wiley and Sons.
Heath, S. 1983. Ways with Words: Language, Life and Work in Communities and Classrooms.
Cambridge: Cambridge University Press.
Heidelberger Forschungsprojekt ‘Pidgin Deutsch’. 1978. ‘The acquisition of German syntax by
foreign migrant workers’ in Sankoff (ed.) 1978.
Hendrickson, J. 1978. ‘Error correction in foreign language teaching: recent theory, research,
and practice’. Modern Language Journal 62: 387-98.
Henning, C. (ed.). 1977. Proceedings of the Los Angeles Second Language Research Forum. Uni¬
versity of California at Los Angeles.
Henzl, V. 1973. ‘Linguistic register of foreign language instruction’. Language Learning 23:
207-27.
Henzl, V. 1979. ‘Foreigner talk in the classroom’. International Review of Applied Linguistics
17: 159-65.
Herbert, R. 1989. ‘The ethnography of compliments’ in Oleksy (ed.) 1989.
Hermann, G. 1980. ‘Attitudes and success in children’s learning of English as a second language:
the motivational vs. the resultative hypothesis’. English Language Teaching Journal 34:
247-54.
 Bibliography 757

Heyde, A. 1979. The relationship between self-esteem and oral production of a second lan¬
guage . Unpublished PhD thesis, Ann Arbor: University of Michigan. Cited in Larsen-
Freeman and Long (eds.) 1991.
Higgs, T. (ed.). 1982. Curriculum, Competence and the Foreign Language Teacher. Skokie, Ili-
nois: National Textbook Company.
Higgs, T. and R. Clifford. 1982. ‘The push toward communication’ in Higgs (ed.) 1982.
Hilles, S. 1986. ‘Interlanguage and the pro-drop parameter’. Second Language Research 2:
33-52.
Hilles, S. 1991. ‘Access to Universal Grammar in second language acquisition’ in Eubank (ed.)
1991b.
Hinde, R. (ed.). 1972. Non-verbal Communication. Cambridge: Cambridge University Press.
Hines, M. and W. Rutherford (eds.). 1981. On TESOL ’81. Washington D.C.: TESOL.
Hirakawa, M. 1989. ‘A study of the L2 acquisition of English reflexives’. Second Language
Research 6: 60-85.
Hirvonen, T. 1985. ‘Children’s foreigner talk: peer talk in play context’ in Gass and Madden
(eds.) 1985.
Holec, H. 1980. ‘Learner training: meeting needs in self-directed learning’ in Altman and
Vaughan James (eds.) 1980.
Holec, H. 1987. ‘The learner as manager: managing learning or managing to learn?’ in Wenden
and Rubin (eds.) 1987.
Holmen, A. 1985. ‘Analysis of some discourse areas in the PIF data and in classroom interaction’
in Glahn and Holmen (eds.) 1985.
Holmes, J. 1986. ‘Compliments and compliment responses in New Zealand English’. Anthrop¬
ological Linguistics 28: 485-508.
Holmes, J. 1988. ‘Paying compliments: a sex-preferential positive politeness strategy’ .Journal of
Pragmatics 12: 445-65.
Holobrow, N., F. Genesee, and W. Lambert. 1991. ‘The effectiveness of a foreign language im¬
mersion program for children from different ethnic and social class backgrounds: Report 2’.
Applied Psycholinguistics 12: 179-98.
Holobrow, N., W. Lambert, and L. Sayegh. 1984. ‘Pairing script and dialogue: combinations
that show promise for second language learning’. Language Learning 34: 59-76.
Homstein, N. and D. Lightfoot (eds.). 1981. Explanation in Linguistics: the Logical Problem of
Language Acquisition. London: Longman.
Horwitz, E. 1986. ‘Preliminary evidence for the reliability and validity of a foreign language
anxiety scale’. TESOL Quarterly 20: 559-62.
Horwitz, E. 1987a. ‘Surveying student beliefs about language learning’ in Wenden and Rubin
(eds.) 1987.
Horwitz, E. 1987b. ‘Linguistic and communicative competence: reassessing foreign language
aptitude’ in VanPatten et al. (eds.) 1987.
Horwitz, E., M. Horwitz, and J. Cope. 1986. ‘Foreign language classroom anxiety’. Modern
Language Journal 70; 125-32.
Horwitz, E. and D. Young. 1991. Language Learning Anxiety: from Theory and Research to
Classroom Implications. Englewood Cliffs, N.J.: Prentice Hall.
Hosenfeld, C. 1978. ‘Students’ mini-theories of second language learning’. Association Bulletin
29:2.
Hosenfeld, C. 1979. ‘Cindy: a learner in today’s foreign language classroom’ in Born (ed.) 1979.
House, J. 1986. ‘Learning to talk: talking to learn. An investigation of learner performance in
two types of discourse’ in Kasper (ed.) 1986.
House, J. and G. Kasper. 1987. ‘Interlanguage pragmatics: requesting in a foreign language’ in
Lorscher and Schultze (eds.) 1987.
Huang, J. 1970. ‘A Chinese child’s acquisition of syntax’. Unpublished MA TESL thesis, Univer¬
sity of California at Los Angeles.
758 Bibliography

Huang, J. and E. Hatch. 1978. ‘A Chinese child’s acquisition of English’ in Hatch (ed.) 1978a.
Huang, X. and M. Van Naerssen. 1985. ‘Learning strategies for oral communication’. Applied
Linguistics 6: 287-307.
Huebner, T. 1979. ‘Order-of-acquisition vs. dynamic paradigm: a comparison of method in in¬
terlanguage research’. TESOL Quarterly 13: 21-8.
Huebner, T. 1980. ‘Creative construction and the case of the misguided pattern’ in Fisher et al.
(eds.) 1980.
Huebner, T. 1983. A Longitudinal Analysis of the Acquisition of English. Ann Arbor: Karoma
Publishers.
Huebner, T. 1985. ‘System and variability in interlanguage syntax’. Language Learning 35:
141-63.
Huebner, T. and C. Ferguson (eds.). 1991. Crosscurrents in Second Language Acquisition and
Linguistic Theories. Amsterdam: John Benjamins.
Hughes, A. and C. Lascaratou. 1982. ‘Competing criteria for error gravity’. English Language
Teaching Journal 36: 175-82.
Hulstijn, J. 1987. ‘Onset and development of grammatical features: two approaches to acquisi¬
tion orders’. Paper given at Interlanguage Conference, Trobe University, Melbourne.
Hulstijn, J. 1989a. ‘A cognitive view of interlanguage variability’ in Eisenstein (ed.) 1989.
Hulstijn, J. 1989b. ‘Implicit and incidental second language learning: experiments in the pro¬
cessing of natural and partly artificial input’ in Dechert and Raupach (eds.) 1989b.
Hulstijn, J. 1990. ‘A comparison between the information-processing and the analysis/control
approaches to language learning’. Applied Linguistics 11: 30-45.
Hulstijn, J. and W. Hulstijn. 1984. ‘Grammatical errors as a function of processing constraints
and explicit knowledge’. Language Learning 34: 23-43.
Hulstijn, J. and E. Marchena. 1989. ‘Avoidance: Grammatical or semantic causes’. Studies in
Second Language Acquisition 11: 242-55.
Hyams, N. 1983. ‘The acquisition of parameterized grammars’. Unpublished PhD thesis, City
University of New York.
Hyams, N. 1991. ‘Seven not-so-trivial trivia of language acquisition: comments on Wolfgang
Klein’ in Eubank (ed.) 1991b.
Hyltenstam, K. 1977. ‘Implicational patterns in interlanguage syntax variation’. Language
Learning 27: 383-411.
Hyltenstam, K. 1984. ‘The use of typological markedness conditions as predictors in second lan¬
guage acquisition: the case of pronominal copies in relative clauses’ in Andersen (ed.)
1984b.
Hyltenstam, K. 1985. ‘L2 learners’ variable output and language teaching’ in Hyltenstam and
Pienemann (eds.) 1985.
Hyltenstam, K. 1990. ‘Typological markedness as a research tool in the study of second language
, acquisition’ in Dechert (ed.) 1990.
Hyltenstam, K. and M. Linnarud (eds.). 1979. Interlanguage Workshop at the Fifth Scandin¬
avian Conference of Linguistics.
Hyltenstam, K. and L. Obler (eds.). 1989. Bilingualism Across the Lifespan: Aspects of Acquisi¬
tion, Maturity and Loss. Cambridge: Cambridge University Press.
Hyltenstam, K. and M. Pienemann (eds.). 1985. Modelling and Assessing Second Language Ac¬
quisition. Clevedon, Avon: Multilingual Matters.
Hymes, D. 1971a. On Communicative Competence. Philadelphia, P.A.: University of Pennsyl¬
vania Press.
Hymes, D. (ed.). 1971b. Pidginization and Creolization of Languages. Cambridge: Cambridge
University Press.

Ioup, G. 1977. ‘Interference versus structural complexity as a predictor of second language

relative clause acquisition’ in Henning (ed.) 1977.
 Bibliography 759

Ioup, G. 1989. ‘Immigrant children who have failed to acquire native English’ in Gass et al (eds 1
1989b.
Ioup, G. and S. Weinberger (eds.). 1987. Interlanguage Phonology: the Acquisition of a Second
Language Sound System. Rowley, Mass.: Newbury House.
Issidorides, D. and J. Hulstijn. 1992. ‘Comprehension of grammatically modified and nonmodi-
fied sentences by second language learners’. Applied Psycholinguistics 13:147-71.
Itoh, H. and E. Hatch. 1978. ‘Second language acquisition: a case study’ in Hatch (ed.) 1978a.

Jackson, H. 1981. ‘Contrastive analysis as a predictor of errors, with reference to Punjabi learn¬
ers of English’ in Fisiak (ed.) 1981b.
Jackson, K. and R. Whitnam. 1971. ‘Evaluation of the Predictive Power of Contrastive Analyses
of Japanese and English’. Final report; Contract No. CEC-0-70-5046 (-823), US Office of
Health, Education and Welfare.
Jaeggli, O. and N. Hyams. 1988. ‘Morphological uniformity and the setting of the null subject
parameter’ in Proceedings of the Northeastern Linguistics Society 18, University of Mas¬
sachusetts at Amherst: Graduate Linguistics Students Association.
Jain, M. 1974. ‘Error analysis: Source, cause and significance’ in Richards (ed.) 1974.
Jakobovits, L. 1970. Foreign Language Learning: a Psycholinguistic Analysis of the Issue. Row-
ley, Mass.: Newbury House.
James, C. 1971. ‘The exculpation of contrastive linguistics’ in Nickel (ed.) 1971.
James, C. 1977. ‘Judgments of error gravity’. English Language Teaching Journal 31: 116-24.
James, C. 1980. Contrastive Analysis. London: Longman.
James, C. 1990. ‘Learner Language’. Language Teaching Abstracts 23: 205-13.
Janicki, K. 1985. The Foreigner’s Language: a Sociolinguistic Perspective. Oxford: Pergamon.
Jarvis G. 1968. ‘A behavioral observation system for classroom foreign language learning’. Mod¬
ern Language Journal 52: 335-41.
Johansson, S. 1973. ‘The identification and evaluation of errors in foreign languages: a func¬
tional approach’ in Svartvik (ed.) 1973a.
Johnson, D. 1992. Approaches to Research in Second Language Learning. New York:
Longman.
Johnson, K. 1988. ‘Mistake correction’. English Language Teaching Journal 42: 89-96.
Johnson, L. and E. Newport. 1989. ‘Critical period effects in second language learning: the influ¬
ence of maturational state on the acquisition of English as a second language’. Cognitive
Psychology 21: 60-99.
Johnson, P. 1981. ‘Effects on reading comprehension of language complexity and cultural back¬
ground of a text’. TESOL Quarterly 15: 169-81.
Johnston, M. 1985. ‘Syntactic and morphological progressions in learner English’. Research re¬
port, Department of Immigration and Ethnic Affairs, Commonwealth of Australia.
Johnston, M. 1986. ‘Second language acquisition research in the adult migrant education pro¬
gram’ in Johnston and Pienemann.
Johnston, M. and M. Pienemann. 1986. Second Language Acquisition: a Classroom Perspective.
New South Wales Migrant Education Service.
Johnston, S. 1990. ‘Teacher questions in the academic language-content classroom’. Unpub¬
lished paper, Tokyo: Temple University Japan.
Jones, A. 1991a. ‘Review of studies on the acquisition of relative clauses in English’. Unpublished
paper, Tokyo: Temple University Japan.
Jones, A. 1991b. ‘Learning to walk before you can run’. Unpublished paper, Tokyo: Temple Uni¬
versity Japan.
Jones, A. 1992. ‘Relatively speaking’. Unpublished paper, Temple University Tokyo.
Jones, F. 1991. ‘Classroom riot: design features, language output and topic in simulations and
other communicative free-stage activities’. System 19:151-69.
Jordens, P. 1980. ‘Interlanguage research: interpretation and explanation’. Language Learning
30: 195-207.
760 Bibliography

Jordens, P. 1988. ‘The acquisition of word order in L2 Dutch and German’ in Jordens and Lalle-
man (eds.) 1988.
Jordens, P. and J. Halleman (eds.). 1988. Language Development. Dordrecht: Foris.
Judd, E. 1978. ‘Language policy and TESOL: socio-political factors and their influences on the
profession’ in Blatchford and Schachter (eds.) 1978.

Kachru, B. 1981. ‘The pragmatics of non-native varieties of English’ in Smith (ed.) 1981.
Kachru, B. 1982. The Other Tongue: English Across Cultures. Urbana Illinois: University of
Illinois Press.
Kachru, B. 1986. The Alchemy of English. Oxford: Pergamon.
Kachru, B. 1989. ‘Teaching World Englishes’. Cross Currents 16: 15-21.
Kadia, K. 1988. ‘The effect of formal instruction on monitored and spontaneous naturalistic in¬
terlanguage performance’. TESOL Quarterly 22: 509-15.
Kamimoto, T., A. Shimura, and E. Kellerman. 1992. ‘A second language classic recon¬
sidered—the case of Schachter’s avoidance’. Second Language Research 8: 231-77.
Kaneko, T. 1991. ‘The role of the LI in second language classrooms’. Unpublished EdD thesis.
Tokyo: Temple University Japan.
Karmiloff-Smith, A. 1986. ‘From metaprocess to conscious access: evidence from children’s
metalinguistic and repair data’. Cognition 28: 95-147.
Kasper, G. 1981. Pragmatische Aspeckte in der Interimsprache. Tubingen: Gunter Narr.
Kasper, G. 1984a. ‘Pragmatic comprehension in learner-native speaker discourse’. Language
Learning 34:1-20.
Kasper, G. 1984b. ‘Perspectives on language transfer’. BAAL Newsletter 24.
Kasper, G. 1985. ‘Repair in foreign language teaching’. Studies in Second Language Acquisition
7: 200-15.
Kasper, G. (ed.). 1986. Learning, Teaching and Communication in the Foreign Language Class¬
room. Aarhus: Aarhus University Press.
Kasper, G. 1989. ‘Variation in speech act realisation’ in Phillipson et al. (eds.) 1991.
Kasper, G. 1990. ‘Linguistic politeness: current research issues’. Journal of Pragmatics 14:
193-218.
Kasper, G. 1992. ‘Pragmatic transfer’. Second Language Research 8: 203-31.
Kasper, G. (ed.). 1988. ‘Classroom Research’. AILA Review 5: 73-88.
Kasper, G. and M. Dahl. 1991. ‘Research methods in interlanguage pragmatics’. Studies in
Second Language Acquisition 12: 215-47.
Kearsley, G. 1976. ‘Questions and question-asking in verbal discourse: a cross-disciplinary
review’. Journal of Pyscholinguistic Research 5: 355-75.
Keefe. J. 1979a. ‘Learning style: an overview’ in Keefe (ed.) 1979b.
Keefe, J. (ed.). 1979b. Student Learning Styles: Diagnosing and Describing Programs. Reston
V.A.: National Secondary School Principals.
Keenan, E. 1974. ‘Conversational competence in children’. Journal of Child Language 1-
163-83.
Keenan, E. 1975. ‘Variation in universal grammar’ in Fasold and Shuy (eds.) 1975.
Keenan, E. and B. Comrie. 1977. ‘Noun phrase accessibility and universal grammar’. Linguistic
Inquiry 8: 63-99.
Kelch, K. 1985. Modified input as an aid to comprehension’. Studies in Second Language Ac¬
quisition 7: 81-90.
Keller, J. 1984. ‘Motivational design of instruction’ in Reigeluth (ed.) 1984.
Kellerman, E. 1977. ‘Towards a characterization of the strategies of transfer in second language
learning’. Interlanguage Studies Bulletin 2: 58-145.
Kellerman, E. 1978. ‘Giving learners a break: native language intuitions as a source of predic¬
tions about transferability’. Working Papers on Bilingualism 15: 59-92.
 Bibliography 761

Kellerman, E. 1979. ‘Transfer and non-transfer: where are we now?’ Studies in Second Language
Acquisition 2: 37-57.
Kellerman, E. 1983. ‘Now you see it, now you don’t’ in Gass and Selinker (eds.) 1983.
Kellerman, E. 1985a. ‘If at first you do succeed ...’ in Gass and Madden (eds.) 1985.
Kellerman, E. 1985b. ‘Dative alternation and the analysis of data: a reply to Mazurkewich’.
Language Learning 35: 91-106.
Kellerman, E. 1986. ‘An eye for an eye: crosslinguistic constraints on the development of the L2
lexicon’ in Kellerman and Sharwood Smith (eds.) 1986.
Kellerman, E. 1987. Aspects of transferability in second language acquisition. Chapter 1: Cross-
linguistic influence: a review. Unpublished manuscript, University of Nijmegen.
Kellerman, E. 1989. ‘The imperfect conditional’ in Hyltenstam and Obler (eds.) 1989.
Kellerman, E. 1991. ‘Compensatory strategies in second language research: a critique, a revision,
and some (non-) implications for the classroom’ in Phillipson et al. (eds.) 1991.
Kellerman, E. 1992. ‘Another look at an old classic; Schachter’s avoidance’. Lecture notes,
Tokyo: Temple University Japan.
Kellerman, E., A. Amerlaan, T. Bongaerts, and N. Poulisse. 1990. ‘System and hierarchy in L2
compensatory strategies’ in Scarcella et al. (eds.) 1990.
Kellerman, E., T. Bongaerts, and N. Poulisse. 1987. ‘Strategy and system in L2 referential com¬
munication’ in Ellis (ed.) 1987a.
Kellerman, E. and M. Sharwood Smith, (eds.). 1986. Cross-linguistic Influence in Second Lan¬
guage Acquisition. Oxford: Pergamon.
Kelley, P. 1982. ‘Interlanguage, variation and sociaL/psychological influences within a develop¬
mental stage’. Unpublished MA in TESL thesis. University of California at Los Angeles.
Ketterman, B. and R. St. Clair (eds.). 1980. New Approaches to Language Acquisition. Heidel¬
berg: Julius Groos.
Khalil, A. 1985. ‘Communicative error evaluations: native speakers’ evaluation and interpreta¬
tion of written errors of Arab EFL learners’. TESOL Quarterly 19: 225-351.
Kilborn, K. 1987. ‘Sentence processing in a second language: seeking a performance definition of
fluency’. Unpublished thesis. University of California, San Diego.
Kilborn, K. and A. Cooremann. 1987. ‘Sentence interpretation strategies in adult Dutch-English
bilinguals’. Applied Psycholinguistics 8: 415-31.
Kilborn, K. and T. Ito. 1989. ‘Sentence processing strategies in adult bilinguals’ in MacWhinney
and Bates (eds.) 1989.
Kinsella, V. (ed.). 1982. Surveys 1: Eight State-of-the-art Articles on Key Areas in Language
Teaching. Cambridge: Cambridge University Press.
Klaus, G. and M. Buhr (eds.). 1976. Philosophisches Worterbuch. Leipzig: VEB Biblio-
graphisches Institut. Cited in Faerch and Kasper 1983.
Kleifgen, J. 1985. ‘Skilled variation in a kindergarten teacher’s use of foreigner talk’ in Gass and
Madden (eds.) 1985.
Klein, W. 1981. ‘Some rules of regular ellipsis in German’ in Klein and Levelt (eds.). 1981
Klein, W. 1986. Second Language Acquisition. Cambridge: Cambridge University Press.
Klein, W. 1990. ‘A theory of language acquisition is not so easy’. Studies in Second Language Ac¬
quisition 12:219-31.
Klein, W. 1991. ‘Seven trivia of language acquisition.’ in Eubank (ed.) 1991b.
Klein, W. and N. Dittmar. 1979. Developing Grammars: the Acquisition of German Syntax by
Foreign Workers. Berlin: Springer.
Klein, W. and W. Levelt (eds.). 1981. Crossing the Boundaries of Linguistics: Studies Presented
to Manfred Bierswisch. Dordrecht: Reidel.
Klein, W. and C. Perdue. 1989. ‘The learner’s problem in arranging words’ in. MacWhinney and
Bates (eds.) 1989.
Klein, W. and B. Rieck. 1982. ‘Der Erwerb der Personalpronomina lm ungesteuerten Spracher-
werb’. Zeitschrift fur Literaturwissenschaft und Linguistik 45: 35-71. Cited in Klein 1986.
762 Bibliography

Kleinmann, H. 1978. ‘The strategy of avoidance in adult second language acquisition’ in Ritchie
(ed.) 1978a.
Klima, E. and V. Bellugi. 1966. ‘Syntactic regularities in the speech of children’ in Lyons and
Wales (eds.) 1966.
Koivukari, A. 1987. ‘Question level and cognitive processing: pyscholinguistic dimensions of
questions and answers’. Applied Psycholinguistics 8:101-20.
Kramarae, C. 1990. ‘Changing the complexion of gender in language research’ in Giles and Rob¬
inson (eds.) 1990.
Kramer, A. 1980. ‘The languages of linguistic theory: aesthetic dimensions of a scientific discip¬
line’. Unpublished doctoral thesis, University of Michigan. Cited in Schumann 1983.
Kramsch, C. 1985. ‘Classroom interaction and discourse options’. Studies in Second Language
Acquisition 7: 169-83.
Kramsch, C. and S. McConnell-Ginet (eds.). 1992. Text and Context: Cross-Disciplinary
Perspectives on Language Study. Lexington, Mass.: D. C. Heath and Company.
Krashen, S. 1973. ‘Lateralization, language learning and the critical period: some new evidence’.
Language Learning 23: 63-74.
Krashen, S. 1976. ‘Formal and informal linguistic environments in language acquisition and lan¬
guage learning’. TESOL Quarterly 10: 157-68.
Krashen, S. 1977. ‘Some issues relating to the Monitor Model’ in H. Brown et al. (eds.) 1977.
Krashen, S. 1978. ‘Individual variation in the use of the monitor’ in Ritchie (ed.) 1978a.
Krashen, S. 1980. ‘The theoretical and practical relevance of simple codes in second language ac¬
quisition’ in Scarcella and Krashen (eds.) 1980.
Krashen, S. 1981. Second Language Acquisition and Second Language Learning. Oxford:
Pergamon.
Krashen, S. 1982. Principles and Practice in Second Language Acquisition. Oxford: Pergamon.
Krashen, S. 1983. ‘Newmark’s ignorance hypothesis and current second language acquisition
theory’ in Gass and Selinker (eds.) 1983.
Krashen, S. 1985. The Input Hypothesis: Issues and Implications. London: Longman.
Krashen, S. 1989. ‘We acquire vocabulary and spelling by reading: additional evidence for the
input hypothesis’. Modern Language Journal 73: 440-64.
Krashen, S., J. Butler, R. Birnbaum, and J. Robertson. 1978. ‘Two studies in language acquisi¬
tion and language learning’. ITL: Review of Applied Linguistics 39/40: 73-92.
Krashen, S., C. Jones, S. Zelinksi, and C. Usprich. 1978. ‘How important is instruction?’ English
Language Teaching Journal 32: 257-61.
Krashen, S., M. Long, and R. Scarcella. 1979. ‘Age, rate and eventual attainment in second lan¬
guage acquisition’. TESOL Quarterly 13: 573-82. Reprinted in Krashen et al. (eds.) 1982.
Krashen, S., M. Long, and R. Scarcella (eds.). 1982. Child-adult Differences in Second Language
Acquisition. Rowley, Mass.: Newbury House.
Krashen, S. and R. Scarcella. 1978. ‘On routines and patterns in second language acquisition and
performance’. Language Learning 28: 283-300.
Krashen, S. and H. Seliger. 1975. ‘The essential characteristics of formal instruction’. TESOL
Quarterly 9: 173-83.
Krashen, S. and H. Seliger. 1976. ‘The role of formal and informal linguistic environments in
adult second language learning’. International Journal of Psycholinguistics 3: 15-21.
Krashen, S., H. Seliger, and D. Hartnett. 1974. ‘Two studies in second language learning’. Kriti-
kon Litterarum 3: 220-8.
Krashen, S. and T. Terrell. 1983. The Natural Approach: Language Acquisition in the Class¬
room. Oxford: Pergamon.
Kruidenier, B. and R. Clement. 1986. ‘The effect of context on the composition and role of ori¬
entations in second language acquisition’. Quebec: International Centre for Research on
Bilingualism.
 Bibliography 763

Kucczaj, S. 1977. ‘Old and new forms, old and new meanings: the form-function hypothesis re¬
visited’. Paper presented at the Society for Research in Child Development, New Orleans.
Kumpf, L. 1984. ‘Temporal systems and universality in interlanguage: a case study’ in Eckman et
al. (eds.) 1984.

Labov, W. 1969a. ‘Contraction, deletion, and inherent variability of the English copula’. Lan¬
guage 45:715-52.
Labov, W. 1969b. ‘The Logic of Non-standard English’. Georgetown Monographs on Language
and Linguistics 22. Washington D.C.: Georgetown University, Center for Applied
Linguistics.
Labov, W. 1970. ‘The study of language in its social context’. Studium Generate 23: 30-87.
Labov, W. 1972. Sociolinguistic Patterns. Philadelphia, P.A.: University of Pennsylvania Press.
Labov, W. 1991. ‘The intersection of sex and social class in the course of linguistic change’. Lan¬
guage Variation and Linguistic Change 2: 205-51.
Lado, R. 1957. Linguistics Across Cultures: Applied Linguistics for Language Teachers. Ann
Arbor, Michigan: University of Michigan.
Lakshmanan, U. 1986. ‘The role of parametric variation in adult second language acquisition: a
study of the “pro-drop” parameter’. Papers in Applied Linguistics—Michigan 2: 97-117.
Lakshmanan, U. 1991. ‘Morphological uniformity and null subjects in child second language
acquisition’ in Eubank (ed.) 1991b.
Lalonde, R. and R. Gardner. 1985. ‘On the predictive validity of the Attitude/Motivation Test
Battery’. Journal of Multilingual and Multicultural Development 6: 403-12.
Lambert, W. 1974. ‘Culture and language as factors in learning and education’ in F. Aboud and
Meade (eds.) 1974.
Lambert, W. and G. Tucker. 1972. The Bilingual Education of Children: the St. Lambert Experi¬
ment. Rowley, Mass.: Newbury House.
Lambert, W., R. Hodgson, R. Gardner, and S. Fillenbaum. 1960. ‘Evaluational reactions to
spoken languages’. Journal of Abnormal and Social Psychology 60: 44-51.
Lane, H. and E. Zale (eds.). 1967. Studies in Language and Language Behaviour, 4.
Langacker, R. 1987. Foundations of Cognitive Grammar. Stanford, California: Stanford
University Press.
Lange, D. 1979. ‘Negation in natiirlichen Englisch-Deutschen Zweitsprachenerwerb: eine
Fallstudie’. International Review of Applied Linguistics 17: 331—48.
Lanoue, G. 1991. ‘Language loss, language gain: cultural camouflage and social change among
the Sekani of Northern British Columbia’. Language in Society 20: 87-115.
Lantolf, J. and A. Labarca (eds.). 1987. Research in Second Language Learning: Focus on the
Classroom. Norwood, N.J.: Ablex.
Larsen-Freeman, D. 1975. ‘The acquisition of grammatical morphemes by adult ESL students’.
TESOL Quarterly 9: 409-30.
Larsen-Freeman, D. 1976a. ‘Teacher speech as input to the ESL learner’. University of Cali¬
fornia. Working Papers in TESL 10: 45-9.
Larsen-Freeman, D. 1976b. ‘An explanation for the morpheme acquisition order of second lan¬
guage learners’. Language Learning 26: 125-34.
Larsen-Freeman, D. 1978. ‘An ESL index of development’. TESOL Quarterly 12: 439-48.
Larsen-Freeman, D. (ed.) 1980. Discourse Analysis in Second Language Research. Rowley,
Mass.: Newbury House.
Larsen-Freeman, D. 1983. ‘The importance of input in second language acquisition’ in Andersen
(ed.) 1983c.
Larsen-Freeman, D. 1991. ‘Second language acquisition research: staking out the territory’.
TESOL Quarterly 25: 315-50.
Larsen-Freeman, D. and M. Long. 1991. An Introduction to Second Language Acquisition Re¬
search. London: Longman.
764 Bibliography

Laudan, L. 1990. Science and Relativism. Some Key Controversies in the Philosophy of Science.
Chicago: University of Chicago Press.
Leather, J. and A. James. 1991. ‘The acquisition of second language speech’. Studies in Second
Language Acquisition 13: 305-41.
Lee, W. 1957. ‘The linguistic context of language learning’. English Language Teaching Journal
11:77-85.
Lee, W. 1968. ‘Thoughts on contrastive linguistics in the context of language teaching’ in Alatis
(ed.) 1968.
Leech, G. 1983. Principles of Pragmatics. London: Longman.
Legaretta, D. 1977. ‘Language choice in bilingual classrooms’. TESOL Quarterly 11: 9-16.
Leino, A. 1982. ‘Learning process in terms of styles and strategies’. Research Bulletin No. 59.
Helsinki, Finland.
Lenneberg, E. 1967. Biological Foundations of Language. New York: Wiley and Sons.
Lennon, P. 1989. ‘Introspection and intentionality in advanced second-language acquisition’.
Language Learning 39: 375-95.
Lennon, P. 1991. ‘Error: Some problems of definition, identification and distinction’. Applied
Linguistics 12: 180-95.
Le Page, R. and A. Tabouret-Keller. 1985. Acts of Identity: Creole-based Approaches to Lan¬
guage and Ethnicity. Cambridge: Cambridge University Press.
Levelt, W. 1983. ‘Monitoring and self-repair in speech’. Cognition 14: 41-104.
Levelt, W. 1989. Speaking: From Intention to Articulation. Cambridge: Cambridge University
Press.
Levenston, E. 1971. ‘Over-indulgence and under-representation: aspects of mother tongue inter¬
ference’ in Nickel (ed.) 1971.
Levenston, E. 1979. ‘Second language lexical acquisition: issues and problems’. Interlanguage
Studies Bulletin 4: 147-60.
Levinson, S. 1983. Principles of Pragmatics. Cambridge: Cambridge University Press.
Li, X. 1989. ‘Effects of contextual cues on inferring and remembering meanings of new words’.
Applied Linguistics 10: 402-13.
Liceras, J. 1985. ‘The role of intake in the determination of learners’ competence’ in Gass and
Madden (eds.) 1985.
Lightbown, P. 1980. ‘The acquisition and use of questions by French L2 learners’ in Felix (ed.)
1980a.
Lightbown, P. 1983. ‘Exploring relationships between developmental and instructional se¬
quences in L2 acquisition’ in Seliger and Long (eds.) 1983.
Lightbown, P. 1984. ‘The relationship between theory and method in second language acquisi¬
tion research’ in Davies et al. (eds.) 1984.
Lightbown, P. 1985a. ‘Can language acquisition be altered by instruction?’ in Hyltenstam and
Pienemann (eds.) 1985.
Lightbown, P. 1985b. ‘Great expectations: second-language acquisition research and classroom
teaching’. Applied Linguistics 6: 173-89.
Lightbown, P. 1987. ‘Classroom language as input to second language acquisition’ in C. Pfaff
(ed.) 1987b.
Lightbown, P. 1990. ‘Process-product research on second language learning in classrooms’ in
Harley et al. (eds.) 1990.
Lightbown, P. 1991a. ‘Getting quality input in the second/foreign language classroom’ in
Kramsch and McConnell-Ginet (eds.) 1991.
Lightbown, P. 1991b. ‘What have we here? Some observations on the effect of instruction on L2
learning’ in Phillipson et al. (eds.) 1991.
Lightbown, P. 1992. ‘Can they do it themselves? A comprehension-based ESL course for young
children’ in Courchene et al. (eds.) 1992.
 Bibliography 765

Lightbown, P. and A. d’Anglejan. 1985. ‘Some input considerations for word order in French LI
and L2 acquisition’ in Gass and Madden (eds.) 1985.
Lightbown, P. and N. Spada. 1990. ‘Focus-on-form and corrective feedback in communicative
language teaching: effects on second language learning’. Studies in Second Language Ac¬
quisition 12: 429^18.
Lightbown, P. and N. Spada. 1993. How Languages are Learned. Oxford: Oxford University
Press.
Lightbown, P., N. Spada, and R. Wallace. 1980. ‘Some effects of instruction on child and adoles¬
cent ESL learners’ in Scarcella and Krashen (eds.) 1980.
Lightbown, P. and L. White. 1987. ‘The influence of linguistic theories on language acquisition
research: description and explanation’. Language Learning 37:483-510.
Little, D. and D. Singleton. 1990. ‘Cognitive style and learning approach’ in Duda and Riley
(eds.) 1990.
Little, D., D. Singleton, and W. Silvius. 1984. Learning Second Languages in Ireland: Experi¬
ence, Attitudes and Needs. Dublin: Trinity College, Centre for Language and Communica¬
tion Studies. Cited in Little and Singleton 1990.
Lococo, V. 1976. ‘A comparison of three methods for the collection of L2 data: free composition,
translation and picture description’. Working Papers on Bilingualism 8: 59-86.
Long, M. 1977. ‘Teacher feedback on learner error: mapping cognitions’ in Brown et al. (eds.)
1977.
Long, M. 1980a. ‘Input, interaction and second language acquisition’. Unpublished PhD thesis.
University of California at Los Angeles.
Long, M. 1980b. ‘Inside the “black box”: methodological issues in classroom research on lan¬
guage learning’. Language Learning 30: 1-42.
Long, M. 1981a. ‘Input, interaction and second language acquisition’ in Winitz (ed.) 1981b.
Long, M. 1981b. ‘Questions in foreigner talk discourse’. Language Learning 31: 135-57.
Long, M. 1983a. ‘Native speaker/non-native speaker conversation and the negotiation of com¬
prehensible input’. Applied Linguistics 4:126^-1.
Long, M. 1983b. ‘Native speaker/non-native speaker conversation in the second language class¬
room’ in Clarke and Handscombe (eds.) 1983.
Long, M. 1983c. ‘Does second language instruction make a difference? A review of the research’.
TESOL Quarterly 17: 359-82.
Long, M. 1984. ‘Process and product in ESL program evaluation’. TESOL Quarterly 18:
409-25.
Long, M. 1985a. ‘Input and second language acquisition theory’ in Gass and Madden (eds.)
1985.
Long, M. 1985b. ‘A role for instruction in second language acquisition: task-based language
teaching’ in Hyltenstam and Pienemann (eds.) 1985.
Long, M. 1985c. ‘Bibliography of research on second language classroom processes and class¬
room second language acquisition’. Technical Report No. 2. Honolulu: Center for Second
Language Classroom Research, Social Science Research Institute, University of Hawaii at
Manoa.
Long, M. 1988. ‘Instructed interlanguage development’ in Beebe (ed.) 1988a.
Long, M. 1989. ‘Task, group, and task-group interactions’. University of Hawaii Working
Papers in ESL 8: 1-26.
Long, M. 1990a. ‘Maturational constraints on language development’. Studies in Second Lan¬
guage Acquisition 12: 251-86.
Long, M. 1990b. ‘Second language classroom research and teacher education’ in Brumfit and
Mitchell (eds.) 1990.
Long, M. 1990c. ‘The least a second language acquisition theory needs to explain’. TESOL
Quarterly 24: 649-66.
766 Bibliography

Long, M. 1991. ‘Focus on form: a design feature in language teaching methodology’ in de Bot et
al. (eds.) 1991.
Long, M. 1993. ‘Assessment strategies for SLA theories’. Applied Linguistics 14: 225-49.
Long, M., L. Adams, M. Mclean, and F. Castanos. 1976. ‘Doing things with words: verbal inter¬
action in lockstep and small group classroom situations’ in Fanselow and Crymes (eds.)
1976.
Long, M. and G. Crookes. 1987. ‘Intervention points in second language classroom processes’ in
Das (ed.) 1987a.
Long, M. and G. Crookes. 1992. ‘Three approaches to task-based syllabus design’. TESOL
Quarterly 26: 27-56.
Long, M., S. Ghambiar, V. Ghambiar, and M. Nishimura. 1982. ‘Regularization in foreigner
talk and interlanguage’. Paper presented at the 17th Annual TESOL Convention, Toronto,
Canada.
Long, M. and P. Porter. 1985. ‘Group work, interlanguage talk, and second language acquisi¬
tion’. TESOL Quarterly 19: 207-28.
Long, M. and C. Sato. 1984. ‘Methodological issues in interlanguage studies: An interactionist
perspective’ in Davies et al. (eds.) 1984.
Lorscher, W. 1986. ‘Conversational structures in the foreign language classroom’ in Kasper (ed.)
1986.
Lorscher, W. and R. Schultze (eds.). 1987. Perspectives on Language in Performance. Tubingen:
Gunter Narr.
Loschky, L. 1989. ‘The effects of negotiated interaction and premodified input on second lan¬
guage comprehension and retention’. Unpublished MA thesis. University of Flawaii.
Loschky, L. and R. Bley-Vroman. 1990. ‘Creating structure-based communication tasks for se¬
cond language development’. University of Flawaii Working Papers in ESL 9:161-212.
Lott, D. 1983. ‘Analysing and counteracting interference errors’. English Language Teaching
Journal 37: 256-61.
Lowenberg, P. 1986. ‘Non-native varieties of English: nativization, norms, and implications’.
Studies in Second Language Acquisition 8: 1-18.
Lucas, E. 1975. ‘Teachers’ reacting moves following errors made by pupils in post-primary Eng¬
lish as a second language classes in Israel’. MA thesis, School of Education, Tel Aviv Univer¬
sity. Cited in Chaudron 1988a.
Ludwig, J. 1982. ‘Native-speaker judgements of second-language learners’ efforts at commun¬
ication: a review’. Modern Language Journal 66: 274-83.
Ludwig, J. 1983. ‘Attitudes and expectations: a profile of female and male students of college
French, German and Spanish’. The Modern Language Journal 67: 216-27.
Lugton, R. (ed.). 1971. Toward a Cognitive Approach to Second Language Acquisition. Phila¬
delphia, Penn.: Center for Curriculum Department.
Lujan, M., L. Minaya, and D. Dankoff. 1984. ‘The universal consistency hypothesis and the pre¬
diction of word order acquisition stages in the speech of bilingual children’. Language 60:
343-71.
Lukmani, Y. 1972. ‘Motivation to learn and language proficiency’. Language Learning 22:
261-73.
Lynch, A. 1988. ‘Speaking up or talking down: foreign learners’ reactions to teacher talk’. Eng¬
lish Language Teaching Journal 42: 109-16.
Lyons, J. 1968. Introduction to Theoretical Linguistics. Cambridge: Cambridge University
Press.
Lyons, J. 1972. ‘Human language’ in Hinde (ed.) 1972.
Lyons, J. and R. Wales (eds.). 1979. Psycholinguistic Papers. Edinburgh: Edinburgh University
Press.
 Bibliography 767

MacIntyre, P. and R. Gardner. 1991a. Methods and results in the study of foreign language
anxiety: a review of the literature’. Language Learning 41: 25-57.
MacIntyre, P. and R. Gardner. 1991b. Language anxiety: its relationship to other anxieties and
to processing in native and second languages. Language Learning 41: 513-34.
MacWhinney, B. (ed.). 1987. Mechanisms of Language Acquisition. Hillsdale, N.J.: Lawrence
Erlbaum.
MacWhinney, B. 1989. ‘Competition and connectionism’ in MacWhinney and Bates (eds.).
MacWhinney, B. and E. Bates (eds.). 1989. The Crosslinguistic Study of Sentence Processing.
Cambridge: Cambridge University Press.
MacWhinney. B., E. Bates, and R. Kligell. 1984. ‘Cue validity and sentence interpretation in
English, German and Italian’. Journal of Verbal Learning and Verbal Behavior 23:127-50.
MacWhinney, B., C. Pleh, and E. Bates. 1985. ‘The development of sentence interpretation in
Hungarian’. Cognitive Psychology 17:178-209.
Magnet, J. 1990. ‘Canadian perspectives on official English’ in Adams and Brink (eds.) 1990.
Maier, M. and P. Jacob. 1966. ‘The effect of variations in self-instruction programs on instruc¬
tional outcomes’. Pyschological Reports 18: 539—46.
Major, R. 1986. ‘Paragoge and degree of foreign accent in Brazilian English’. Second Language
Research 2: 53-71.
Major, R. 1987. ‘A model for interlanguage pronunciation’ in Ioup and Weinberger (eds.) 1987.
Major, J. 1988. ‘Balancing form and function’. International Review of Applied Linguistics 26:
81-100.
Makino, T. 1980. ‘Acquisition order of English morphemes by Japanese secondary school
students’. Journal of Hokkaido University Education 30: 101-48.
Malouf, R. and C. Dodd. 1972. ‘Role of exposure, imitation and expansion in the acquisition of
an artificial grammatical rule’. Developmental Psychology 7: 195-203.
Maltz, D. and R. Borker. 1982. ‘A cultural approach to male-female miscommunication’ in
Gumperz (ed.) 1982b.
Mangubhai, F. 1991. ‘The processing behaviours of adult second language learners and their re¬
lationship to second language proficiency’. Applied Linguistics 12: 268-98.
Mannon, T. 1986. ‘Teacher talk: a comparison of a teacher’s speech to native and non-native
speakers’. Unpublished MA TESL thesis, University of California at Los Angeles.
Maple, R. 1982. ‘Social distance and the acquisition of English as a second language: a study of
Spanish-speaking adult learners’. Unpublished doctoral thesis, University of Texas at
Austin.
Martin, G. 1980. ‘English language acquisition: the effects of living with an American family’.
TESOL Quarterly 14: 388-90.
Mason, C. 1971. ‘The relevance of intensive training in English as a foreign language for univer¬
sity students’. Language Learning 21: 197-204.
Mateene, K. and J. Kalema (eds.). 1980. ‘Reconsideration of African linguistic policies’. OAU/
BIL Publication 3. Kampala: OAU Bureau of Languages.
Mateene, K., J. Kalema, and B. Chomba (eds.). 1985. ‘Linguistic liberation and unity of Africa’.
OAU/BIL Publication 6. Kampala: OAU Bureau of Languages
Matsumoto, K. 1989. ‘Factors involved in the L2 learning process’. JALTJournal 11: 167-92.
Mazurkewich, 1.1985. ‘Syntactic markedness and language acquisition’. Studies in Second Lan¬
guage Acquisition 7: 15-35.
Mazurkewich, I. 1984. ‘The acquisition of dative alternation by second language learners and
linguistic theory’. Language Learning 34: 91-109.
McCary, J. (ed.). 1956. Psychology of Personality. New York: Logos.
McClelland, J., D. Rumelhart, and G. Hinton. 1986. ‘The appeal of parallel distributed pro¬
cessing’ in Rumelhart et al. (eds.) 1986.
768 Bibliography

McClelland, J., D. Rumelhart, and the PDP Research Group (eds.). 1986. Parallel Distributed
Processing: Explorations in the Microstructure of Cognition, Vol. 2: Psychological and Bio¬
logical Models. Cambridge, Mass.: MIT Press.
McDonald, F., M. Stone, and A. Yates. 1977. ‘The effects of classroom interaction patterns and
student characteristics on the acquisition of proficiency in English as a second language .
Princeton, N.J.: Educational Testing Service.
McDonald, J. 1986. ‘The development of sentence comprehension strategies in English and
Dutch’. Journal of Experimental Child Psychology 41: 317-35.
McDonald, J. and K. Heilenman. 1991. ‘Determinants of cue strength in adult first and second
language speakers of French’. Applied Psycholinguistics 12: 313-48.
McHoul, A. 1978. ‘The organization of turns at formal talk in the classroom’. Language and
Society 7: 183-213.
McLaughlin, B. 1978a. Second Language Acquisition in Childhood. Hillsdale, N.J.: Lawrence
Erlbaum.
McLaughlin, B. 1978b. ‘The Monitor Model: some methodological considerations’. Language
Learning 28: 309-32.
McLaughlin, B. 1980. ‘Theory and research in second-language learning: an emerging para¬
digm’. Language Learning 30: 331-50.
McLaughlin, B. 1985. Second Language Acquisition in Childhood, Vol. 2: School-age Children.
Hillsdale, N.J.: Lawrence Erlbaum.
McLaughlin, B. 1987. Theories of Second Language Learning. London: Edward Arnold.
McLaughlin, B. 1990a. ‘Restructuring’. Applied Linguistics 11: 113-28.
McLaughlin, B. 1990b. ‘The relationship between first and second languages: language profi¬
ciency and language aptitude’ Harley et al. (eds.) 1990.
McLaughlin, B. 1990c. ‘ “Conscious” vs. “unconscious” learning’. TESOL Quarterly 24:
617-34.
McLaughlin, B. and M. Harrington. 1989. ‘Second-language acquisition’. Annual Review of
Applied Linguistics 10: 122-34.
McLaughlin, B., T. Rossman, and B. McLeod. 1983. ‘Second language learning: an information¬
processing perspective’. Language Learning 33: 135-58.
McNamara, J. 1973. ‘Nurseries, streets and classrooms: some comparisons and deductions’.
Modern Language Journal 57: 250-55.
McNeill, D. 1966. ‘Developmental psycholinguistics’ in Smith and Miller (eds.) 1966.
McNeill, D. 1970. The Acquisition of Language. New York: Harper Row.
McTear, M. 1975. ‘Structure and categories of foreign language teaching sequences’ in All-
wright(ed.) 1975b.
Meara, P. 1984. ‘The study of lexis in interlanguage’ in Davies et al. (eds.) 1984.
Meara, P. (ed.). 1986. Spoken Language. London: CILT.
Mehan, H. 1974. ‘Accomplishing classroom lessons’ in A. Cicourel et al. (eds.) 1974.
Mehan, H. 1979. Learning Lessons: Social Organization in the Classroom. Cambridge, Mass.:
Harvard University Press.
Meisel, J. 1980. ‘Linguistic simplification’ in Felix (ed.) 1980a.
Meisel, J. 1983. ‘Strategies of second language acquisition: more than one kind of simplification’
in Andersen (ed.) 1983c.
Meisel, J. 1987. ‘Reference to past events and actions in the development of natural second lan¬
guage acquisition’ in Pfaff (ed.) 1987b.
Meisel, J. 1991. ‘Principles of Universal Grammar and strategies of language learning: some
similarities and differences between first and second language acquisition’ in Eubank (ed.)
1991b.
Meisel, J., H. Clahsen, and M. Pienemann. 1981. ‘On determining developmental stages in
natural second language acquisition’. Studies in Second Language Acquisition 3: 109-35.
 Bibliography 7 69

Midorikawa, H. 1990. ‘An exploratory study of the relationship of teacher questions to learner
output’. Unpublished paper, Tokyo: Temple University Japan.
Miller, G, andN. Chomsky. 1963. ‘Finitary models of language users’ in Bush et al. (eds.) 1963.
Milon, J. 1974. ‘The development of negation in English by a second language learner’. TESOL
Quarterly 8: 137-43.
Mitchell, R. 1985. ‘Process research in second language classrooms’. Language Teaching 18:
330-52.
Mitchell, R., B. Parkinson, and R. Johnstone. 1981. ‘The foreign language classroom: an obser¬
vational study’. Stirling Educational Monographs No. 9. Stirling: Department of Educa¬
tion, University of Stirling.
Mizon, S. 1981. ‘Teacher talk: a case study from the Bangalore/Madras communicational ELT
project’. Unpublished MA thesis. University of Lancaster.
Mohle, D. and M. Raupach. 1989. ‘Language transfer and procedural knowledge’ in Dechert
and Raupach (eds.) 1989a.
Montgomery, C. and M. Eisenstein. 1985. ‘Real reality revisited: an experimental communica¬
tive course in ESL’. TESOL Quarterly 19: 317—33.
Moore, T. (ed.). 1973. Cognitive Development and the Acquisition of Language. New York:
Academic Press.
Morren, B. 1993. ‘Lexical inferencing procedures (K. Haastrup 1989)—a replication’. Unpub¬
lished paper, Tokyo: Temple University Japan.
Morris, R. (ed.). 1989. Parallel Distributed Processing: Implications for Psychology and Neuro¬
biology. Oxford: Clarendon Press.
Morrison, D. and G. Low. 1983. ‘Monitoring and the second language learner’ in Richards and
Schmidt (eds.) 1983.
Moskowitz, G. 1967. ‘The Flint system: an observational tool for the foreign language class¬
room’ in. Simon and Boyer (eds.) 1967.
Moskowitz, G. 1978. Caring and Sharing in the Poreign Language Classroom. Rowley, Mass.:
Newbury House.
Muchnick, A. and D. Wolfe. 1982. ‘Attitudes and motivations of American students of Spanish’.
Canadian Modern Language Review 38:262-81.
Miihlhauser, P. 1986. Pidgin and Creole Linguistics. Oxford: Basil Blackwell.
Mukkatesh, L. 1977. Problematic areas in English syntax for Jordanian students. University of
Amman, Jordan.
Mukkatesh. L. 1986. ‘Persistence in fossilization’. International Review of Applied Linguistics
24:187-203.
Muysken, P. 1984. The Spanish that Quechua speakers learn’ in Andersen (ed.) 1984b.
Myers, T. <ed. . 1979. The Development of Conversation and Discourse. Edinburgh: Edinburgh
University Press.

Nagara, S. 1972. Japanese Pidgin English in Hawaii: a Bilingual Description. Honolulu: Univer¬
sity Press of Hawaii.
Naiman, N., M. Frohlich, H. Stern, and A. Todesco. 1978. The Good Language Learner. Re¬
search in Education Series No 7. Toronto: The Ontario Institute for Studies in Education.
Naro, A. 1983. ‘Comments on “Simplified Input and Second Language Acquisition” ’ in Ander¬
sen (ed. i 1983c. ^
Nation, P. 1990. Teaching and Learning Vocabulary. New York: Newbury House/Harper Row.
Nation, R. and B. McLaughlin. 1986. ‘Experts and novices: an information-processing ap¬
proach to the “good language learner” problem’. Aprplied Pyscholinguistics 1: 41-56.
Nattinger. J. and J. DeCarrico. 1992. Lexical Phrases and Language Teaching. Oxford: Oxford
University Press.
770 Bibliography

Nayak, N., N. Hansen, N. Krueger, and B. McLaughlin. 1987. ‘Language-learning strategies in

monolingual and multilingual subjects’. Unpublished paper, Santa Cruz: University of Cali¬
fornia. Cited in McLaughlin 1990b.
Nelson, K. 1973. ‘Structure and Strategy in Learning to Talk’. Monographs of the Society for Re¬
search in Child Development 38.
Nelson, K. 1977. ‘Facilitating children’s syntax acquisition’. Developmental Psychology 13:
101-7.
Nelson, K. (ed.). 1978. Children’s Language, Vol 1. New York: Gardner’s Press.
Nelson, K. (ed.). 1980. Children’s Language, Vol 2. New York: Gardner’s Press.
Nemser, W. 1971. ‘Approximative systems of foreign language learners’. International Review
of Applied Linguistics 9: 115-23.
Neu, J., and S. Gass (eds.). (forthcoming). Speech Acts Across Cultures. Berlin: Mouton de
Gruyter.
Neufeld, G. 1977. ‘Language learning ability in adults: a study on the acquisition of prosodic and
articulatory features’. Working Papers on Bilingualism 12: 45-60.
Neufeld, G. 1978. ‘On the acquisition of prosodic and articulatory features in adult language
learning’. Canadian Modern Language Review 34:163-74.
Neufeld, G. 1979. ‘Toward a theory of language learning ability’. Language Learning 29:
227-41.
Newmark, L. 1966. ‘How not to interfere in language learning’. International Journal of Amer¬
ican Linguistics 32: 77-87.
Newmark, L. and D. Reibel. 1968. ‘Necessity and sufficiency in language learning’. Interna¬
tional Review of Applied Linguistics in Language Teaching 6: 145-64.
Newport, E. 1976. ‘Motherese: the speech of mothers to young children’. Unpublished PhD the¬
sis, University of Pennsylvania.
Newport, E., H. Gleitman, and L. Gleitman. 1977. ‘ “Mother, I’d rather do it myself”: some ef¬
fects and non-effects of maternal speech styles’ in Snow and Ferguson (eds.) 1977.
Newton, J. 1991. ‘Negotiation: negotiating what?’ Paper given at SEAMEO Conference on Lan¬
guage Acquisition and the Second/Foreign Language Classroom, RELC, Singapore.
Neymeyer, F. (ed.). 1988. Linguistics: The Cambridge Survey, Vol IV: The Socio-cultural Con¬
text. Cambridge: Cambridge University Press.
Nicholas, H. 1985. ‘Learner variation and the teachability hypothesis’ in Hyltenstam and Piene-
mann (eds.) 1985.
Nicholas, H. 1986. ‘The acquisition of language as the acquisition of variation’. Australian
Working Papers on Language Development 1: 1-30.
Nickel, G. (ed.). 1971. Papers in Contrastive Analysis. Cambridge: Cambridge University Press.
Nickel, G. 1973. ‘Aspects of error analysis and grading’ in Svartvik (ed.) 1973a.
Nobuyoshi, J. and R. Ellis. 1993. ‘Focused communication tasks’. English Language Teaching
Journal 47:203-10.
Nunan, D. 1989. Designing tasks for the communicative classroom. Cambridge: Cambridge
University Press.
Nunan, D. 1990a. ‘The questions teachers ask’. JALTJournal 12:187-202.
Nunan, D. 1990b. ‘The teacher as researcher’ in Brumfit and Mitchell (eds.) 1990.
Nunan, D. 1991. ‘Methods in second language classroom-oriented research: a critical review’.
Studies in Second Language Acquisition 13: 249-74.
Nyikos, M. 1990. ‘Sex related differences in adult language learning: socialization and memory
factors’. Modern Language Journal 3: 273-87.
Nystrom, N. 1983. ‘Teacher-student interaction in bilingual classrooms: four approaches to
error feedback’ in Seliger and Long (eds.) 1983.

Obler, L. 1989. ‘Exceptional second language learners’ in Gass et al. (eds.) 1989b.
Ochs, E. 1979. ‘Planned and unplanned discourse’ in Givon (ed.) 1979b.
 Bibliography 771

Ochs, E. 1982. ‘Talking to children in Western Samoa’. Language in Society 11: 77-104.
Ochs, E. and B. Schiefflin (eds.). 1979. Developmental Pragmatics. New York: Academic Press.
Ochs, E. and B. Schieffelin. 1984. ‘Language acquisition and socialization: three developmental
stories and their implications’ in Shweder and LeVine (eds.) 1984.
Odlin, T. 1989. Language Transfer. Cambridge: Cambridge University Press.
Odlin, T. 1990. ‘Word-order transfer, metalinguistic awareness and constraints on foreign lan¬
guage learning’ in VanPatten and Lee (eds.) 1990.
O’Grady, W. 1991. ‘Language acquisition and the “pro-drop” phenomenon: a response to
Hilles’ in Eubank (ed.) 1991b.
Okamura-Bichard, F. 1985. ‘Mother tongue maintenance and second language learning: a case
of Japanese children’. Language Learning 35: 63-89.
Oleksy, W. (ed).) 1989. Contrastive Pragmatics. Amsterdam: John Benjamins.
Oiler, J. 1977. ‘Attitude variables in second language acquisition’ in Burt et al. (eds.) 1977.
Oiler, J. 1981. ‘Research on the measurement of affective variables: some remaining questions’
in Andersen (ed.) 1981.
Oiler, J. (ed.). 1983. Issues in Language Testing Research. Rowley, Mass.: Newbury House.
Oiler, J., L. Baca, and A. Vigil. 1977. ‘Attitudes and attained proficiency in ESL: a sociolinguistic
study of Mexican Americans in the Southwest’. TESOL Quarterly 11: 173-83.
Oiler, J. and K. Perkins. 1978. ‘Intelligence and language proficiency as sources of variance in
self-reported affective variables’. Language Learning 28: 85-97.
Olsen, L. and S. Samuels. 1973. ‘The relationship between age and accuracy of foreign language
pronunciation’. Journal of Educational Research 66: 263-67. Reprinted in Krashen et al.
(eds.) 1979.
Olshtain, E. 1983. ‘Sociocultural competence and language transfer: the case of apologies’ in
Gass and Selinker (eds.) 1983.
Olshtain, E. 1989. ‘Apologies across languages’ in Blum-Kulka et al. (eds.) 1989b.
Olshtain, E. and S. Blum-Kulka. 1985. ‘Degree of approximation: non-native reactions to native
speech act behavior’ in Gass and Madden (eds.) 1985.
Olshtain, E. and A. Cohen. 1983. ‘Apology: a speech act set’ in Wolfson and Judd (eds.) 1983.
Olshtain, E. and A. Cohen. 1989. ‘Speech act behavior across languages’ in Dechert and Rau-
pach (eds.) 1989a.
Olshtain, E., E. Shohamy, J. Kemp, and R. Chatow. 1990. ‘Factors predicting success in EFL
among culturally different learners’. Language Learning 40: 23M4.
Olshtain, E. and L. Weinbach. 1985. ‘Complaints: a study of speech act behavior among native
and non-native speakers of Hebrew’ in Verschuren and Bertucelli-Papi (eds.) 1985.
O’Malley, J. 1987. ‘The effects of training in the use of learning strategies on acquiring English as
a second language’ in Wenden and Rubin (eds.) 1987.
O’Malley, J. and A. Chamot. 1990. Learning Strategies in Second Language Acquisition. Cam¬
bridge: Cambridge University Press.
O’Malley, J., A. Chamot, G. Stewner-Manzanaraes, L. Kupper, and R. Russo. 1985a. ‘Learning
strategies used by beginning and intermediate ESL students’. Language Learning 35:21—46.
O’Malley, J., A. Chamot, G. Stewner-Manzanaraes, R. Russo, and L. Kupper. 1985b. ‘Learning
strategy applications with students of English as a second language’. TESOL Quarterly 19:
285-96.
O’Malley, J., A. Chamot, and C. Walker. 1987. ‘Some applications of cognitive theory to second
language acquisition’. Studies in Second Language Acquisition 9: 287-306.
Oxford, R. 1985. ‘A new taxonomy of second language learning strategies’. Washington D.C.:
ERIC Clearinghouse on Languages and Linguistics.
Oxford, R. 1986. ‘Development of the strategy inventory for language learning’. Manuscript.
Washington D.C.: Center for Applied Linguistics.
Oxford, R. 1989. ‘Use of language learning strategies: a synthesis of studies with implications for
teacher training’. System 17: 235-47.
772 Bibliography

Oxford, R. 1990. Language Learning Strategies: What Every Teacher Should Know. Rowley,
Mass.: Newbury House.
Oxford, R. 1992. ‘Who are our students? A synthesis of foreign and second language research on
individual differences with implications for instructional practice’. TESL Canada Journal 9:
30-19.
Oxford, R. and M. Nyikos. 1989. ‘Variables affecting choice of language learning strategies by
university students’. Modern Language Journal 73: 291-300.
Oyama, S. 1976. ‘A sensitive period in the acquisition of a non-native phonological system’.
Journal of Psych olinguistic Research 5: 261-85.

Palmberg, R. (ed.). 1979. Perception and Production of English: Papers on Interlanguage.

AFTIL Vol 6. Publications of the Dept, of English, Abo Akademi.
Palmer, A. 1979. ‘Compartmentalized and integrated control: an assessment of some evidence
for two kinds of competence and implications for the classroom’. Language Learning 29:
169-80.
Pankhurst, J., M. Sharwood Smith, and P. Van Buren (eds.). 1988. Learnability and Second Lan¬
guages: a Book of Readings. Dordrecht: Foris.
Paribakht, T. 1985. ‘Strategic competence and language proficiency’. Applied Linguistics 6:
132-46.
Parker, K. 1989. ‘Learnability theory and the acquisition of syntax’. University of Hawaii.
Working Papers in ESL 8: 49-78.
Parker, K. and C. Chaudron. 1987. ‘The effects of linguistic simplifications and elaborative mo¬
difications on L2 comprehension’. University of Hawaii. Working Papers in ESL 6:107-33.
Parkinson, B. and C. Howell-Richardson. 1990. ‘Learner Diaries’ in Brumfit and Mitchell (eds.)
1990.
Parry, T. and J. Child. 1990. ‘Preliminary investigation of the relationship between VORD,
MLAT and language proficiency’ in Parry and Stansfield (eds.).
Parry, T. and C. Stansfield. (ed.). 1990. Language Aptitude Reconsidered. Englewood Cliffs,
N.J.: Prentice Hall.
Pask, G. and B. Scott. 1972. ‘Learning strategies and individual competence’. International
Journal of Man-Machine Studies 4: 217-53.
Patkowski, M. 1980. ‘The sensitive period for the acquisition of syntax in a second language’.
Language Learning 30: 449-72.
Patkowski, M. 1990. ‘Age and accent in a second language: a reply to James Emir Flege’. Applied
Linguistics 11: 73-89.
Pavesi, M. 1984. ‘The acquisition of relative clauses in a formal and informal context’ in Single-
ton and Little (eds.) 1984.
Pavesi, M. 1986. ‘Markedness, discoursal modes and relative clause formation in a formal and
informal context’. Studies in Second Language Acquisition 8: 38-55.
Pawley, A. and F. Syder. 1983. ‘Two puzzles for linguistic theory: nativelike selection and nat¬
ivelike fluency’ in Richards and Schmidt (eds.) 1983.
Peck, S. 1978. ‘Child-child discourse in second language acquisition’ in Hatch (ed.) 1978a.
Peck, S. 1980. ‘Language play in child second language acquisition’ in Larsen-Freeman (ed.)
1980.
Pelligrini, A. and T. Yawkey (eds.). 1984. The Development of Oral and Written Language in
Social Contexts. Norwood, N.J.: Ablex.
Penfield, W. and L. Roberts. 1959. Speech and Brain Mechanisms. New York: Atheneum Press.
Perdue, C. 1984. Second Language Acquisition by Adult Immigrants: a Field Manual. Rowley,
Mass.: Newbury House.
Perdue, C. 1991. ‘Cross-linguistic comparisons: organizational principles in learner languages’
in Huebner and Ferguson (eds.) 1991.
 Bibliography 773

Perdue, C. and W. Klein. 1992. ‘Why does the production of some learners not grammaticalize?’
Studies in Second Language Acquisition 14: 259-72.
Perkins, K. and D. Larsen-Freeman. 1975. ‘The effects of formal language instruction on the or¬
der of morpheme acquisition’. Language Learning 25: 237-43.
Perren, G. and J. Trim (eds.). 1971. Applications of Linguistics. Cambridge University Press.
Peters, A. 1977. ‘Language learning strategies: does the whole equal the sum of the parts?’ Lan¬
guage 53: 560-73. Also in Diller (ed.) 1981.
Petersen, C. and A. Al-Haik. 1976. ‘The development of the Defense Language Aptitude Battery
(DLAB)’. Educational and Psychological Measurement 36: 369-80.
Pfaff, C. 1987a. ‘Functional approaches to interlanguage’ in Pfaff (ed.).
Pfaff, C. (ed.). 1987b. First and Second Language Acquisition Processes. Cambridge, Mass.:
Newbury House.
Pfaff, C. 1992. ‘The issue of grammaticalization in early German second language’. Studies in
Second Language Acquisition 14: 273-96.
Phillips, S. 1972. ‘Participation structures and communicative competence: Warm Springs chil¬
dren in community and classroom’ in Cazden et al. (eds.) 1972.
Phillipson, R., E. Kellerman, L. Selinker, M. Sharwood Smith, and M. Swain (eds.). 1991. For¬
eign/Second Language Pedagogy Research. Clevedon, Avon: Multilingual Matters.
Phinney, M. 1987. ‘The pro-drop parameter in second language acquisition’ in Roeper and Wil¬
liams (eds.) 1987.
Piatelli-Palmarini, M. (ed.). 1980. Language and Learning. London: Routledge and Kegan Paul.
Pica, T. 1983. ‘Adult acquisition of English as a second language under different conditions of
exposure’. Language Learning 33: 465-97.
Pica, T. 1984. ‘Methods of morpheme quantification: their effect on the interpretation of second
language data’. Studies in Second Language Acquisition 6: 69-78.
Pica, T. 1985. ‘The selective impact of instruction on second language acquisition’. Applied Lin¬
guistics 6:214-22.
Pica, T. 1987. ‘Second language acquisition, social interaction in the classroom’. Applied Lin¬
guistics 7: 1-25.
Pica, T. 1988. ‘Interlanguage adjustments as an outcome of NS-NNS negotiated interaction’.
Language Learning 38: 45-73.
Pica, T. 1991. ‘Classroom interaction, participation and comprehension: redefining relation¬
ships’. System 19: 437-52.
Pica, T. 1992. ‘The textual outcomes of native speaker-non-native speaker negotiation: what do
they reveal about second language learning’ in Kramsch and McConnell-Ginet (eds.) 1992.
Pica, T. and C. Doughty. 1985a. ‘Input and interaction in the communicative language class¬
room: a comparison of teacher-fronted and group activities’ in Gass and Madden (eds.)
1985.
Pica, T. and C. Doughty. 1985b. ‘The role of group work in classroom second language acquisi¬
tion’. Studies in Second Language Acquisition 7: 233-48.
Pica, T. and C. Doughty. 1988. ‘Variations in classroom interaction as a function of participa¬
tion pattern and task’ in Fine (ed.) 1988.
Pica, T., L. Holliday, N. Lewis, D. Berducci, and J. Newman. 1991. ‘Language learning through
interaction: what role does gender play?’ Studies in Second Language Acquisition 13:
343-76.
Pica, T., L. Holliday, N. Lewis, and L. Morgenthaler. 1989. ‘Comprehensible output as an out¬
come of linguistic demands on the learner’. Studies in Second Language Acquisition 11:
63-90.
Pica, T., R. Kanagy, and J. Falodun (forthcoming). ‘Choosing and using communication tasks
for second language research and instruction’ in. Gass and Crookes (eds.).
Pica, T. and M. Long. 1986. ‘The linguistic and conversational performance of experienced and
inexperienced teachers’ in Day (ed.) 1986.
774 Bibliography

Pica, T., R. Young, and C. Doughty. 1987. ‘The impact of interaction on comprehension’.
TESOL Quarterly 21: 737-58.
Pickett, G. 1978. The Foreign Language Learning Process. London: The British Council.
Pienemann, M. 1980. ‘The second language acquisition of immigrant children’ in Felix (ed.)
1980a.
Pienemann, M. 1981. Der Zweitsprachenerwerb auslandischer Arbeitskinder. Bonn: Bouvier.
Pienemann, M. 1984. ‘Psychological constraints on the teachability of languages’. Studies in
Second Language Acquisition 6: 186-214.
Pienemann, M. 1985. ‘Learnability and syllabus construction’ in Hyltenstam and Pienemann
(eds.) 1985.
Pienemann, M. 1986. ‘Is language teachable? Psycholinguistic experiments and hypotheses’.
Australian Working Papers in Language Development 1: 3.
Pienemann, M. 1987. ‘Determining the influence of instruction on L2 speech processing’.
Australian Review of Applied Linguistics 10: 83-113.
Pienemann, M. 1989. ‘Is language teachable? Psycholinguistic experiments and hypotheses’.
Applied Linguistics 10: 52-79.
Pienemann, M. 1992. ‘COALA—A computational system for interlanguage analysis’. Second
Language Research 8: 59-92.
Pienemann, M. and M. Johnston. 1986. ‘An acquisition-based procedure for second language
assessment’. Australian Review of Applied Linguistics 9: 92-122.
Pienemann, M. and M. Johnston. 1988. ‘Processing constraints and learnability’. Unpublished
manuscript, University of Sydney.
Pienemann, M., M. Johnston, and G. Brindley. 1988. ‘Constructing an acquisition-based pro¬
cedure for assessing second language acquisition’. Studies in Second Language Acquisition
10:217-43.
Pimsleur, P. 1966. Pimsleur Language Aptitude Battery (PLAB). New York: Harcourt Brace
Jovanovich.
Pimsleur, P. and T. Quinn (eds.). 1971. The Psychology of Second Language Learning. Cam¬
bridge; Cambridge University Press.
Pimsleur, P., D. Sundland, and R. MacIntyre. 1966. Underachievement in Foreign Language
Learning. Washington D.C.: Modern Language Association.
Pinker, S. and A. Prince. 1989. ‘Rules and connections in human language’ in Morris (ed.) 1989.
Plann, S. 1977. ‘Acquiring a second language in an immersion situation’ in H. Brown et al. (eds.)
1977.
Pleh, C. 1990. ‘The search for universal operating principles in language acquisition’. Studies in
Second Language Acquisition 12: 233-41.
Plunkett, K. 1988. ‘Parallel distributed processing’. Psyke and Logos 9,2: 307-36.
Politzer, R. 1970. ‘Some reflections on “good” and “bad” language teaching behaviors’. Lan¬
guage Learning 20: 31—43.
Politzer, R. 1980. ‘Requesting in elementary school classrooms’. TESOL Quarterly 14:165-74.
Politzer, R. 1983. ‘An exploratory study of self-reported language learning behaviors and their
relation to achievement’. Studies in Second Language Acquisition 6: 54-67.
Politzer, R. and M. McGroarty. 1985. ‘An exploratory study of learning behaviors and their re¬
lationship to gains in linguistic and communicative competence’. TESOL Quarterly
19:103-23.
Politzer, R. and A. Ramirez. 1973. ‘An error analysis of the spoken English of Mexican-
American pupils in a bilingual school and a monolingual school’. Language Learning 18:
35-53.
Politzer, R., A. Ramirez, and S. Lewis. 1981. ‘Teaching standard English in the third grade: class¬
room functions of language’. Language Learning SI-. 171-93.
Poole, D. 1992. ‘Language socialization in the second language classroom’. Language Learning
42:593-616.
 Bibliography 775

Porter, J. 1977. ‘A cross-sectional study of morpheme acquisition in first language learners’.

Language Learning 27: 47-62.
Porter, P. 1986. ‘How learners talk to each other: input and interaction in task-centred discus¬
sions’ in Day (ed.) 1986.
Poulisse, N. 1990a. The Use of Compensatory Strategies by Dutch Learners of English.
Enschede: Sneldruk.
Poulisse, N. 1990b. ‘Variation in learners’ use of communication strategies’ in Duda and Riley
(eds.) 1990.
Prabhu, N.S. 1987. Second Language Pedagogy. Oxford: Oxford University Press.
Prator, C. 1968. ‘The British heresy in TESL’ in Fishman et al. (eds.) 1968.
Preston, D. 1981. ‘The ethnography of TESOL’. TESOL Quarterly 15: 105-16.
Preston, D. 1989. Sociolinguistics and Second Language Acquisition. Oxford: Blackwell.
Pride, J. (ed.). 1979. Sociolinguistic Aspects of Language Learning and Teaching. Oxford:
Oxford University Press.
Purcell, E. and R. Suter. 1980. ‘Predictors of pronunciation accuracy: a reexamination’. Lan¬
guage Learning 30: 271-87.

Quirk, R. 1982a. ‘International communication and the concept of nuclear English’ in Quirk
(ed.).
Quirk, R. (ed.). 1982b. Style and Communication in the English Language. London: Edward
Arnold.
Quirk, R. 1985. ‘The English language in a global context’ in Quirk and Widdowson (eds.).
Quirk, R. and H. Widdowson (eds.). 1985. English in the World: Teaching and Learning the
Language and Literatures. Cambridge: Cambridge University Press.

Rabinowitz, M. and M. Chi. 1987. ‘An interactive model of strategic processing’ in Ceci (ed.)
1987.
Ramage, K. 1990. ‘Motivational factors and persistence in foreign language study’. Language
Learning 40: 189-219.
Ramat, A. 1992. ‘Grammaticalization processes in the area of temporal and modal relations’.
Studies in Second Language Acquisition 14: 297-322.
Ramirez, A. and N. Stromquist. 1979. ‘ESL methodology and student language learning in bi¬
lingual elementary schools’. TESOL Quarterly 13: 145-58.
Ramirez, J., S. Yuen, D. Ramey, and B. Merino. 1986. ‘First year report: longitudinal study of
immersion programs for language minority children’. Arlington, V.A.: SRA Technologies.
Rampton, B. 1987. ‘Stylistic variability and not speaking “normal” English: some post-
Labovian approaches and their implications for the study of interlanguage’ in Ellis (ed.)
1987a.
Raupach, M. 1983. ‘Analysis and evaluation of communication strategies’ in Fserch and Kasper
(eds.) 1983a.
Ravem, R. 1968. ‘Language acquisition in a second language environment’. International Re¬
view of Applied Linguistics 6: 165-85.
Reber, A. 1976. ‘Implicit learning of synthetic learners: the role of instructional set’. Journal of
Experimental Psychology, Human Learning and Memory 2: 88-94.
Reber, A., S. Kassin, S. Lewis, and G. Cantor. 1980. ‘On the relationship between implicit and
explicit modes in the learning of a complex rule structure’. Journal of Experimental
Psychology, Human Learning and Memory 6: 492-502.
Reed, C. (ed.). 1971. The Learning of Language. New York: Appleton-Century-Crofts.
Reid, J. 1987. ‘The learning style preferences of ESL students’. TESOL Quarterly 21: 87-111.
Reigeluth, C. (ed.). 1984. Instructional Design Theories and Models. Hillsdale, N.J.: Lawrence
Erlbaum. Cited in Crookes and Schmidt 1989.
776 Bibliography

Reiss, M. 1983. ‘Helping the unsuccessful language learner’. The Canadian Modern Language
Review 39: 257-66.
Reiss, M. 1985. ‘The good language learner: another look’. Canadian Modern Language Review
41:511-23.
Rescorla, L. and S. Okuda. 1987. ‘Modular patterns in second language acquisition’. Applied
Psycholinguistics 8: 281-308.
Reynolds, P. 1971. A Primer in Theory Construction. Indianapolis, Bobbs Merrill.
Richards, J. 1971a. ‘Error analysis and second language strategies’. Language Sciences 17:
12-22.
Richards, J. 1971b. ‘A non-contrastive approach to error analysis’. English Language Teaching
Journal 25: 204-19.
Richards, J. (ed.). 1974. Error Analysis. London: Longman.
Richards, J. (ed.). 1978. Understanding Second and Foreign Language Learning: Issues and
Approaches. Rowley, Mass.: Newbury House.
Richards, J. 1972. ‘Social factors, interlanguage and language learning’. Language Learning 22:
159-88.
Richards, J. andD. Nunan (eds.). 1990. Second Language Teacher Education. Cambridge: Cam¬
bridge University Press.
Richards, J. and T. Rogers. 1986. Approaches and Methods in Language Teaching. Cambridge:
Cambridge University Press.
Richards, J. and J. Schmidt (eds.). 1983. Language and Communication. London: Longman.
Riley, P. 1977. ‘Discourse networks in classroom interaction: some problems in communicative
language teaching’. Melanges Pedagogiques, University of Nancy: CRAPEL.
Riley, P. 1985. Discourse and Learning. London: Longman.
Riley, P. 1989. ‘Well don’t blame me! On the interpretation of pragmatic errors’ in Oleksy (ed.)
1989.
Riney, T. 1990. ‘Age and open syllable preference in interlanguage phonology’ in Burmeister and
Rounds (eds.) 1990.
Ringbom, H. 1986. ‘Crosslinguistic influence and the foreign language learning process’ in Kel-
lerman and Sharwood Smith (eds.) 1986.
Ringbom, H. 1978. ‘The influence of the mother tongue on the translation of lexical items’. In¬
terlanguage Studies Bulletin 3: 80-101.
Ringbom, H. 1987. The Role of the First Language in Foreign Language Learning. Clevedon,
Avon: Multilingual Matters.
Ringbom, H. 1992. ‘On LI transfer in L2 comprehension and L2 production’. Language Learn¬
ing 42:85-112.
Ringbom, H. and R. Palmberg (eds.). 1976. Errors made by Finns and Swedish-speaking Finns in
the Learning of English. Abo, Linland: Dept, of English, Abo Akademi, ERIC Report ED
>22628.
Rintell, E. 1981. ‘Sociolinguistic variation and pragmatic ability: a look at learners’. Interna¬
tional Journal of the Sociology of Language 27: 11-34.
Rintell, E. and C. Mitchell. 1989. ‘Studying requests and apologies’ in Blum-Kulka et al. (eds.)
1989a.
Ritchie, W. (ed.). 1978a. Second Language Acquisition Research. New York: Academic Press.
Ritchie, W. 1978b. ‘The right roof constraint in adult-acquired language’ in Ritchie (ed.).
Rivera, C. (ed.). 1984. Communicative Competence Approaches to Language Proficiency As-
sessment: Research and Application. Clevedon, Avon: Multilingual Matters.
Roberts, C. and M. Simonot. 1987. ‘ “This is my life”: how language acquisition is interac-
tionally accomplished’ in Ellis (ed.) 1987a.
Robinett, B. and J. Schachter (eds.). 1983. Second Language Learning: Contrastive Analysis,
Error Analysis and Related Aspects. Ann Arbor: The University of Michigan.
 Bibliography 111

Robson, G. 1992. ‘Individual learner differences and classroom participation: a pilot study’.
Unpublished paper, Tokyo: Temple University Japan.
Robson, G. 1993. ‘The effects of learner personality and anxiety on participation and the rela¬
tionship between participation and proficiency’. Doctoral thesis proposal, Tokyo: Temple
University Japan.
Rodriguez, R. 1982. Hunger of Memory: the Education of Richard Rodriguez. Boston: David R.
Godine.
Roeper, T. and E. Williams (eds.). 1987. Parameter Setting. Dordrecht: Reidel.
Rogers, S. (ed.). 1976. They Don’t Speak Our Language. London: Edward Arnold.
Romaine, S. 1984. The Language of Children and Adolescents. Oxford: Blackwell.
Rosansky, E. 1976. ‘Methods and morphemes in second language acquisition’. Language Learn¬
ing 26:409-25.
Rosenstein, D. 1982. ‘The efficiency of correcting recurrent errors among EFL students’. Beer-
sheva, Israel: Department of English as a Foreign Language, Ben-Gurion University. Cited
in Cohen et al. 1991.
Rosing-Schow, D. and K. Haastrup. 1982. ‘The use of communication strategies in classroom
and spontaneous discourse’. Unpublished manuscript, Department of English, University of
Copenhagen. Cited in Holmen 1985.
Rossier, J. 1975. ‘Extroversion-introversion as a significant variable in the learning of English as
a second language’. Unpublished doctoral thesis, University of Southern California. Disser¬
tation Abstracts International 36: 7308A-7309A.
Rost, M. 1990. Listening in Language Learning. London: Longman.
Rost, M. and S. Ross. 1991. ‘Learner strategies in interaction: typology and teachability’. Lan¬
guage Learning 41: 235—73.
Rubin, J. 1975. ‘What the “good language learner” can teach us’. TESOL Quarterly 9: 41-51.
Rubin, J. 1981. ‘Study of cognitive processes in second language learning’. Applied Linguistics
11: 117-31.
Rubin, J. 1987. ‘Learner strategies: theoretical assumptions, research history and typology’ in
Wenden and Rubin (eds.) 1987.
Rulon, K. and J. McCreary. 1986. ‘Negotiation of content: teacher-fronted and small group in¬
teraction’ in Day (ed.) 1986.
Rumelhart, D. and J. McClelland. 1986. ‘On learning the past tenses of English verbs’ in McClel¬
land et al. (eds.) 1986.
Rumelhart, D., J. McClelland, and the PDP Research Group (eds.). 1986. Parallel Distributed
Processing: Explorations in the Microstructures of Cognition,Vol 2. Psychological and
Biological Models. Cambridge, Mass.: MIT Press.
Rutherford, W. 1982. ‘Markedness in second language acquisition’. Language Learning 32:
85-108.
Rutherford, W. 1983. ‘Language typology and language transfer’ in Gass and Selinker (eds.)
1983.
Rutherford, W. (ed.). 1984a. Typological Universals and Second Language Acquisition. Amster¬
dam: John Benjamins.
Rutherford, W. 1984b. ‘Description and explanation in interlanguage syntax: state of the art’.
Language Learning 34: 127-55.
Rutherford, W. 1988. Second Language Grammar: Learning and Teaching. London: Longman.
Rutherford, W. 1989. ‘Preemption and the learning of L2 grammars’. Studies in Second Lan¬
guage Acquisition 11: 441-57.
Rutherford, W. and M. Sharwood Smith. 1985. ‘Consciousness raising and Universal Gram¬
mar’. Applied Linguistics 6: 274-82.
Rutherford, W. and M. Sharwood Smith (eds.). 1988. Grammar and Second Language Teach¬
ing: A Book of Readings. Rowley, Mass.: Newbury House.
778 Bibliography

Ryan, E. and H. Giles (eds.). 1982. Attitudes Towards Language Variation. London: Edward
Arnold.

Sachs, J. 1977. ‘The adaptive significance of linguistic input to prelinguistic infants’ in Snow and
Ferguson (eds.) 1977.
Sacks, H., E. Schegloff, and G. Jefferson. 1974. ‘A simplest systematics for the organization of
turn taking in conversation’. Language 50: 696-735.
Sadtano, E. (ed.). 1991. Language Acquisition and the Second/Foreign Language Classroom.
Singapore: SEAMEO Regional Language Centre.
Saito, M. 1985. ‘Some asymmetries in Japanese and their theoretical implications’. Unpublished
doctoral thesis, Cambridge, Mass.: MIT Press.
Sajavaara, K. 1981a. ‘Contrastive linguistics past and present and a communicative approach’ in
Fisiak (ed.) 1981b.
Sajavaara, K. 1981b. ‘The nature of first language transfer: English as L2 in a foreign language
setting’. Paper presented at the first European-North American Workshop in Second Lan¬
guage Acquisition Research. Lake Arrowhead, California.
Salica, C. 1981. ‘Testing a model of corrective discourse’. Unpublished MA in TESL thesis. Uni¬
versity of California at Los Angeles. Cited in Chaudron 1988.
Sankoff, D. (ed.). 1978. Linguistic Variation: Model and Methods. New York: Academic Press.
Santos, T. 1987. ‘Markedness theory and error evaluation: an experimental study’. Applied Lin¬
guistics 8:207-18.
Sasaki, Y. 1991. ‘English and Japanese interlanguage comprehension strategies: an analysis
based on the competition model’. Applied Psycholinguistics 12: 47-73.
Sato, C. 1981. ‘Ethnic styles in classroom discourse’ in Hines and Rutherford (eds.) 1981.
Sato, C. 1984. ‘Phonological processes in second language acquisition: another look at in¬
terlanguage syllable structure’. Language Learning 34:43-57. Also in Ioup and Weinberger
(eds.) 1987.
Sato, C. 1985. ‘Task variation in interlanguage phonology’ in Gass and Madden (eds.) 1985.
Sato, C. 1987. ‘Phonological processes in second language acquisition: another look at in¬
terlanguage syllable structure’ in Ioup and Weinberger (eds.) 1987.
Sato, C. 1988. ‘Origins of complex syntax in interlanguage development’. Studies in Second Lan¬
guage Acquisition 10: 371-95.
Saunders, N. 1987. ‘Morphophonemic variations in clusters in Japanese English’. Language
Learning 37: 247-72.
Savignon, S. 1972. Communicative Competence: an Experiment in Foreign Language Teaching.
Philadelphia: Center for Curriculum Development.
Saville-Troike, M. 1988. ‘Private speech: evidence for second language learning strategies during
the “silent period” '.Journal of Child Language 15: 567-90.
Savord, J-G. and L. Laforge (eds.). 1981. Proceedings of the Fifth Conference of AILA. Laval:
University of Laval Press.
Scarcella, R. 1979. ‘On speaking politely in a second language’ in Yorio et al. (eds.) 1979.
Scarcella, R. 1983. ‘Discourse accent in second language performance’ in Gass and Selinker
(eds.) 1983.
Scarcella, R., E. Andersen, and S. Krashen (eds.). 1990. Developing Communicative Compe¬
tence. New York: Newbury House.
Scarcella, R. and C. Higa. 1981. ‘Input, negotiation and age differences in second language ac¬
quisition’. Language Learning 31: 409-37.
Scarcella, R. and S. Krashen. (eds.). 1980. Research in Second Language Acquisition. Rowley,
Mass.: Newbury House.
Schachter, J. 1974. ‘An error in error analysis’. Language Learning 27: 205-14.
Schachter, J. 1983. ‘A new account of language transfer’ in Gass and Selinker (eds.) 1983.
 Bibliography 779

Schachter, J. 1986a. ‘In search of systematicity in interlanguage production’. Studies in Second

Language Acquisition 8:119-34.
Schachter, J. 1986b. ‘Three approaches to the study of input’. Language Learning 36: 211-25.
Schachter, J. 1988. ‘Second language acquisition and its relationship to Universal Grammar’.
Applied Linguistics 9: 219-35.
Schachter, J. 1989. ‘Testing a proposed universal’ in Gass and Schachter (eds.) 1989.
Schachter, J. 1991. ‘Corrective feedback in historical perspective’. Second Language Research 7:
89-102.
Schachter, J. and M. Celce-Murcia. 1977. ‘Some reservations concerning error analysis’. TESOL
Quarterly 11: 441-51.
Schachter, J. and W. Rutherford. 1979. ‘Discourse function and language transfer’. Working
Papers on Bilingualism 19: 3-12.
Schachter, J. and V. Yip. 1990. ‘Grammaticality judgments: why does everyone object to subject
extraction?’ Studies in Second Language Acquisition 12: 379-92.
Schegloff, E., G. Jefferson, and H. Sacks. 1977. ‘The preference for self-correction in the or¬
ganization of repair in conversation’. Language 53: 361-82.
Scherer, A. and M. Wertheimer. 1964. A Psycholinguistic Experiment in Foreign Language
Teaching. New York: McGraw Hill.
Scherer, K. and H. Giles (eds.). 1979. Social Markers in Speech. Cambridge: Cambridge Univer¬
sity Press.
Schinke-Llano, L. 1990. ‘Can foreign language learning be like second language acquisition? The
curious case of immersion’ in VanPatten and Lee (eds.) 1990.
Schmidt, M. 1980. ‘Coordinate structures and language universals in interlanguage’. Language
Learning 30: 397-416.
Schmidt, R. 1977. ‘Sociolinguistic variation and language transfer in phonology’. Working Pa¬
pers on Bilingualism 12: 79-95.
Schmidt, R. 1983. ‘Interaction, acculturation and the acquisition of communication compet¬
ence’ in Wolfson and Judd (eds.) 1983.
Schmidt, R. 1988. ‘The potential of parallel distributed processing for S.L.A. theory and re¬
search’. University of Hawaii Working Papers in ESL 7: 55-56.
Schmidt, R. 1990. ‘The role of consciousness in second language learning’. Applied Linguistics
11: 129-58.
Schmidt, R. 1992. ‘Psychological mechanisms underlying second language fluency’. Studies in
Second Language Acquisition 14: 357-85.
Schmidt, R. and S. Frota. 1986. ‘Developing basic conversational ability in a second language: a
case-study of an adult learner’ in Day (ed.) 1986.
Schouten, M. 1979. ‘The missing data in second language learning research’. Interlanguage
Studies Bulletin 4: 3-14.
Schumann, F. and J. Schumann. 1977. ‘Diary of a language learner: an introspective study of
second language learning’ in H. Brown et al. (eds.) 1977.
Schumann, J. 1978a. ‘The acculturation model for second language acquisition’ in Gmgras (ed.)
1978.
Schumann, J. 1978b. The Pidginization Process: a Model for Second Language Acquisition.
Rowley, Mass.: Newbury House.
Schumann, J. 1978c. ‘Social and psychological factors in second language acquisition’ in
Richards, (ed.) 1978.
Schumann, J. 1979. ‘The acquisition of English negation by speakers of Spanish: a review of the
literature’ in Andersen (ed.) 1979a.
Schumann, J. 1980. ‘The acquisition of English relative clauses by second language learners’ in
Scarcella and Krashen (eds.) 1980.
Schumann, J. 1983. ‘Art and science in second language acquisition research’. Language Learn¬
ing Special Issue 33: 49-75.
780 Bibliography

Schumann, J. 1986. ‘Research on the acculturation model for second language acquisition’.
Journal of Multilingual and Multicultural Development 7: 379-92.
Schumann, J. 1987. ‘The expression of temporality in basilang speech’. Studies in Second Lan¬
guage Acquisition 9: 21-41.
Schumann, J. 1990. ‘Extending the scope of the acculturation/pidginization model to include
cognition’. TESOL Quarterly 24: 667-84.
Schumann, J. 1993. ‘Some problems with falsification: an illustration from SLA research’. Ap¬
plied Linguistics 14: 295-306.
Schumann, J. and N. Stenson. (eds.). 1974. New Frontiers in Second Language Learning. Row-
ley, Mass.: Newbury House.
Schwartz, B. 1986. ‘The epistemological status of second language acquisition’. Second Lan¬
guage Research 2: 120-59.
Schwartz, B. and B. Gubala-Ryzak. 1992. ‘Learnability and grammar reorganization in L2A:
against negative evidence causing the unlearning of verb movement’. Second Language
Research 8: 1-38.
Schwartz, J. 1980. ‘The negotiation for meaning: repair in conversations between second lan¬
guage learners of English’ in Larsen-Freeman (ed.) 1980.
Schwartz, R. and S. Camarata. 1985. ‘Examining the relationship between input and language
development: some statistical issues ’.Journal of Child Language 12: 13-25.
Schweder, R. and R. LeVine (eds.). 1984. Culture and its Acquisition. New York: Academic
Press.
Scollon, R. 1976. Conversations with a One-year Old. Honolulu: The University of Hawaii
Press.
Scollon, R. and S. Scollon. 1983. ‘Face in interethnic communication’ in Richards and Schmidt
(eds.) 1983.
Scovel, T. 1978. ‘The effect of affect on foreign language learning: a review of the anxiety re¬
search’. Language Learning 28: 129-42.
Scovel, T. 1981. ‘The recognition of foreign accents in English and its implications for psycho-
linguistic theories of language acquisition’ in Savord and Laforge (eds.) 1981.
Scovel, T. 1988. A Time to Speak: a Psycholinguistic Enquiry into the Critical Period for Human
Speech. Rowley, Mass.: Newbury House.
Scribner, S. and M. Cole. 1973. ‘Cognitive consequences of formal and informal learning’.
Science 182: 553-9.
Searle, J. 1969. Speech Acts. Cambridge: Cambridge University Press.
Searle, J. 1975. ‘Indirect speech acts’ in Cole and Morgan (eds.) 1975.
Searle, J. 1976. ‘The classification of illocutionary acts’. Language in Society 5:1-24.
Seliger, H. 1975. ‘Inductive method and deductive method in language teaching: a re¬
examination’. International Review of Applied Linguistics 13: 1-18.
Seliger, H. 1977. ‘Does practice make perfect? A study of the interaction patterns and L2 compe¬
tence’. Language Learning 27: 263-78.
Seliger, H. 1978. ‘Implications of a multiple critical periods hypothesis for second language
learning’ in Ritchie (ed.) 1978a.
Seliger, H. 1979. ‘On the nature and function of language rules in language teaching’. TESOL
Quarterly 13: 359-69.
Seliger, H. 1984. ‘Processing universal in second language acquisition’ in Eckman et al. (eds.)
1984.
Seliger, H. 1989. ‘Semantic transfer constraints in foreign language speakers’ reactions to accept¬
ability’ in Dechert and Raupach (eds.) 1989a.
Seliger, H. and M. Long (eds.). 1983. Classroom-oriented Research in Second Language Ac¬
quisition. Rowley, Mass.: Newbury House.
Selinker, L. 1969. ‘Language transfer’. General Linguistics 9: 67-92.
Selinker, L. 1972. ‘Interlanguage’. International Review of Applied Linguistics 10: 209-31.
 Bibliography 781

Selinker, L. 1984. ‘The current state of interlanguage studies: an attempted critical summary’ in
Davies et al. (eds.) 1984.
Selinker, L. 1992. Rediscovering Interlanguage. London: Longman.
Selinker, L. and D. Douglas. 1985. ‘Wrestling with context in interlanguage theory’. Applied
Linguistics 6: 190-204.
Selinker, L. and J. Lamendella. 1978. ‘Two perspectives on fossilization in interlanguage learn¬
ing’. Interlanguage Studies Bulletin 3:143-91.
Selinker, L., M. Swain, and G. Dumas. 1975. ‘The interlanguage hypothesis extended to chil¬
dren’. Language Learning 25: 139-91.
Shapiro, F. 1979. ‘What do teachers actually do in language classrooms?’ Paper presented at the
13th Annual TESOL Convention, Boston.
Shapson, S. and V. D’Oyley (eds.). 1984. Bilingual and Multicultural Education: Canadian Per¬
spectives. Clevedon, Avon: Multilingual Matters.
Sharwood Smith, M. 1980. ‘The competence-performance distinction in the theory of second
language and the Pedagogical Grammar Hypothesis’. Paper presented at the Contrastive
Linguistics Conference, Boszkowo. Cited in Rutherford and Sharwood Smith 1985.
Sharwood Smith, M. 1981. ‘Consciousness-raising and the second language learner’. Applied
Linguistics 2:159-69.
Sharwood Smith, M. 1984. ‘Discussant of “The study of lexis in interlanguage” by Paul Meara’
in Davies et al. (eds.) 1984.
Sharwood Smith, M. 1985. ‘From input to intake: on argumentation in second language acquisi¬
tion’ in Gass and Madden (eds.) 1985.
Sharwood Smith M. 1986. ‘Comprehension vs. acquisition: two ways of processing input’. Ap¬
plied Linguistics 7: 239-56.
Sharwood Smith, M. 1991. ‘Speaking to many minds: on the relevance of different types of lan¬
guage information for the L2 learner’. Second Language Research 7: 118-32.
Sharwood Smith, M. and E. Kellerman. 1986. ‘Crosslinguistic influence in second language ac¬
quisition: an introduction’ in Kellerman and Sharwood Smith (eds.) 1986.
Shatz, M. 1978. ‘On the development of communicative understanding: an early strategy for
interpreting and responding to message’. Journal of Cognitive Psychology 10: 271-301.
Shatz, M. and R. Gelman. 1973. ‘The development of communication skills: modifications in the
speech of young children as a function of listener’. Monographs of the Society for Research
in Child Development 38, no. 5.
Sheorey, R. 1986. ‘Error perceptions of native-speaking and non-native speaking teachers of
ESL’. English Language Teaching Journal 40: 306-12.
Shiffrin, R. and W. Schneider. 1977. ‘Controlled and automatic human information processing:
II. Perceptual learning, automatic attending and a general theory’. Psychological Review 84:
127-90.
Shimura, A. 1990. ‘Why do the Japanese learners produce fewer relative clauses in English’.
Unpublished term paper. ESL 750: University of Hawaii.
Simon, A. and E. Boyer (eds.). 1967. Mirrors for behavior: an anthology of classroom observa¬
tion instruments. Philadelphia: Center for the Study of Teaching at Temple University.
Sinclair, J. and D. Brazil. 1982. Teacher Talk. Oxford: Oxford University Press.
Sinclair, J. and M. Coulthard. 1975. Towards an Analysis of Discourse. Oxford: Oxford Univer¬
sity Press.
Sinclair-de-Zwart, H. 1973. ‘Language acquisition and cognitive development’ in Moore (ed.)
1973.
Singleton, D. 1987. ‘Mother and other tongue influence on learner French: a case study’. Studies
in Second Language Acquisition 9: 327-45.
Singleton, D. 1989. Language Acquisition: the Age Factor. Clevedon, Avon: Multilingual
Matters.
782 Bibliography

Singleton, D. and D. Little (eds.). 1984. Language Learning in Formal and Informal Contexts.
Dublin: IRAAL.
Sjoholm, K. 1976. ‘A comparison of the test results in grammar and vocabulary between Finnish-
and Swedish-speaking applicants for English’ in Ringbom and Palmberg (eds.) 1976.
Sjoholm, K. 1979. ‘Do Finns and Swedish-speaking Finns use different strategies in the learning
of English as a foreign language?’ in Palmberg (ed.) 1979.
Skehan, P. 1986a. ‘Cluster analysis and the identification of learner types’ in Cook (ed.) 1986.
Skehan, P. 1986b. ‘Where does language aptitude come from?’ in Meara (ed.) 1986.
Skehan, P. 1989. Individual Differences in Second-language Learning. London: Edward Arnold.
Skehan, P. 1990. The relationship between native and foreign language learning ability: educa¬
tional and linguistic factors’ in Dechert (ed.) 1990.
Skehan, P. 1991. ‘Individual differences in second language learning’. Studies in Second Lan¬
guage Acquisition 13: 275-98.
Skinner, B. 1957. Verbal Behavior. New York: Appleton-Century-Crofts.
Skuttnab-Kangas, T. (ed.). 1977. Papers from the First Nordic Conference on Bilingualism. Hel¬
singfors: Universitet.
Skuttnab-Kangas, T. 1986. ‘Who wants to change what and why—conflicting paradigms in mi¬
nority education research’ in Spolsky (ed.) 1986.
Skuttnab-Kangas, T. 1988. ‘Multilingualism and the education of minority children’ in Skutt¬
nab-Kangas and Cummins (eds.) 1988.
Skuttnab-Kangas, T. and J. Cummins (eds.). 1988. Minority Education. Clevedon, Avon: Multi¬
lingual Matters.
Slimani, A. 1989. ‘The role of topicalization in classroom language learning’. System 17:
223-34. __
Slobin, D. 1970. ‘Universal of grammatical development in children’ in Flores d’Arcais and
Levelt (eds.) 1970.
Slobin, D. 1971. Psycholinguistics. Glenview, Illinois: Scott Foresman.
Slobin, D. 1973. ‘Cognitive prerequisites for the development of grammar’ in Ferguson and Slo¬
bin (eds.) 1973.
Slobin, D. (ed.). 1985a. The Crosslinguistic Study of Language Acquisition, Vol. 2, Theoretical
Issues. Hillsdale, N.J.: Lawrence Erlbaum.
Slobin, D. 1985b. ‘Cross-linguistic evidence for the language-making capacity’ in Slobin (ed.).
Smith, D. 1972. ‘Some implications for the social status of pidgin languages’ in Smith and Shuy
(eds.) 1972.
Smith, D. and R. Shuy (eds.). 1972. Sociolinguistics in Cross-cultural Analysis. Washington
D.C.: Georgetown University Press.
Smith, F. and G. Miller (eds.). 1966. The Genesis of Language: a Psycholinguistic Approach.
Cambridge, Mass.: MIT Press.
Smith, L. (ed.). 1981. English for Cross-cultural Communication. London: Macmillan.
Smith, L. 1983. Readings in English as an International Language. London: Pergamon.
Smith, P. 1970. ‘A comparison of the audiolingual and cognitive approaches to foreign language
instruction: the Pennsylvania Foreign Language Project’. Philadelphia: Center for Curric¬
ulum Development.
Snow, C. 1972. ‘Mother’s speech to children learning’. Child Development 43: 549-65.
Snow, C. 1976. ‘The language of the mother-child relationship’ in Rogers (ed.) 1976.
Snow, C. 1977. ‘Mothers’ speech research: from input to interaction’ in Snow and Ferguson
(eds.) 1977.
Snow, C. 1986. ‘Conversations with Children’ in Fletcher and Garman (eds.) 1986.
Snow, C. 1987. ‘Beyond conversation: second language learners’ acquisition of description and
explanation’ in Lantolf and Labarca (eds.) 1987.
Snow, C. and C. Ferguson (eds.). 1977. Talking to Children: Language Input and Acquisition.
Cambridge: Cambridge University Press.
 Bibliography 783

* Snow, C. and M. Hoefnagel-Hohle. 1978. The critical age for language acquisition: evidence
from second language learning’. Child Development 49: 1114—28.
Snow, C. and M. Hoefnagel-Hohle. 1982. ‘School age second language learners’ access to sim¬
plified linguistic input’. Language Learning 32: 411-30.
Sokolik, M. 1990. ‘Learning without rules: PDP and a resolution of the adult language learning
paradox’. TESOL Quarterly 24: 685-96.
Sorace, A. 1985. ‘Metalinguistic knowledge and language use in acquisition-poor environ¬
ments’. Applied Linguistics 6: 239-54.
Sorace, A. 1988. ‘Linguistic intuitions in interlanguage development: the problem of indetermin¬
acy’ in Pankhurst et al. (eds.) 1988.
Spada, N. 1986. ‘The interaction between types of content and type of instruction: some effects
on the L2 proficiency of adult learners’. Studies in Second Language Acquisition 8: 181-99.
Spada, N. 1987. ‘Relationships between instructional differences and learning outcomes: a pro¬
cess-product study of communicative language teaching’. Applied Linguistics 8: 137-61.
Spada, N. and P. Lightbown. 1989. ‘Intensive ESL programmes in Quebec primary schools’.
TESL Canada 7:11-32.
Spielberger, C. 1983. Manual for the State-Trait Anxiety Inventory (Form Y). Palo Alto, Calif.:
Consulting Psychologists Press.
Spindler, G. (ed.). 1982. Doing the Ethnography of Schooling: Educational Ethnography in Ac¬
tion. New York: CBS College Publishing. Cited in Mitchell 1985.
Spolsky, B. 1969. ‘Attitudinal aspects of second language learning’. Language Learning 19:
271-85.
Spolsky, B. 1986a. ‘Overcoming language barriers to education in a multilingual world’ in Spol¬
sky (ed.).
Spolsky, B. (ed.). 1986b. Language and Education in Multilingual Settings. Clevedon, Avon:
Multilingual Matters.
Spolsky. B. 1988. ‘Bridging the gap: a general theory of second language learning’. TESOL
Quarterly 22: 377-96.
Spolsky, B. 1989. Conditions for Second Language Learning. Oxford: Oxford University Press.
Spolsky, B. 1990. ‘Introduction to a colloquium: the scope and form of a theory of second lan¬
guage learning’. TESOL Quarterly 24: 609-16.
Sridhar, S. and K. Sridhar. 1986. ‘Bridging the paradigm gap: second language acquisition theory
and indigenized varieties of English’. World Englishes 5: 3-14.
Stansfield, C. and L. Hansen. 1983. ‘Field-dependence-independence as a variable in second lan¬
guage cloze test performance’. TESOL Quarterly 17: 29-38.
Stauble, A. 1978. ‘Decreolization: a model for second language development’. Language Learn¬
ing 28:29-54.
Stauble, A. 1984. ‘A comparison of the Spanish-English and Japanese-English interlanguage
continuum’ in Andersen (ed.) 1984b.
Stenson, B. 1974. ‘Induced errors’ in Schumann and Stenson (eds.) 1974.
Stern, H. 1975. ‘What can we learn from the good language learner?’ Canadian Modern Lan¬
guage Review 31: 304-18.
Stern, H. 1983. Fundamental Concepts of Language Teaching. Oxford: Oxford University
Press.
Stern, H. 1990. ‘Analysis and experience as variables in second language pedagogy’ in Harley et
al. (eds.) 1990.
Stever, S., C. Walker, and S. Mufwene (eds.). 1976. Papers from the Parasession on Diachronic
Syntax. Chicago: Chicago Linguistic Society.
Stevick, E. 1980. Teaching Languages: a Way and Ways. Rowley, Mass.: Newbury House.
Stevick, E. 1989. Success with Foreign Languages. New York: Prentice Hall.
Stockwell, R. and J. Bowen. 1965. The Sounds of English and Spanish. Chicago: Chicago Uni¬
versity Press.
784 Bibliography

Stockwell, R., J. Bowen, and J. Martin. 1965. The Grammatical Structures of English and Span¬
ish. Chicago: Chicago University Press.
Strevens, P. 1980. Teaching English as an International Language. Oxford: Pergamon.
Strong, M. 1983. ‘Social styles and second language acquisition of Spanish-speaking kinder¬
garteners’. TESOL Quarterly 17: 241-58.
Strong, M. 1984. ‘Integrative motivation: cause or result of successful second language acquisi¬
tion?’ Language Learning 34:1-14.
Sundquist, L. 1986. ‘Lexical inferencing among Swedish- and Finnish-speaking primary-school
pupils’. Unpublished Masters thesis. Abo Akademi University, Abo (Turku), Finland.
Svanes, B. 1988. ‘Attitudes and “cultural distance” in second language acquisition’. Applied
Linguistics 9: 357-71.
Svartvik, J. (ed.). 1973a. Errata: Papers in Error Analysis. Lund, Sweden: CWK Gleerup.
Svartvik, J. 1973b. ‘Introduction’ in Svartvik (ed.).
Swain, M. 1981. ‘Target language use in the wider environment as a factor in its acquisition’ in
Andersen (ed.) 1981.
Swain, M. 1985. ‘Communicative competence: some roles of comprehensible input and compre¬
hensible output in its development’ in Gass and Madden (eds.) 1985.
Swain, M. and J. Cummins. 1979. ‘Bilingualism, cognitive functioning and education’. Lan¬
guage Teaching and Abstracts 4-18. Reprinted in Kinsella (ed.) 1982.
Swain, M. and S. Lapkin. 1982. Evaluating Bilingual Education: a Canadian Case Study. Cleve-
don, Avon: Multilingual Matters.
Swan, M. 1987. ‘Non-systematic variability: a self-inflicted conundrum’ in Ellis (ed.) 1987a.

Tahta, S., M. Wood, and K. Loewenthal. 1981. ‘Age changes in the ability to replicate foreign
pronunciation and intonation’. Language and Speech 24: 363-72.
Tajfel, FI. (ed.). 1984. The Social Dimension, Vol 2. Cambridge, Cambridge University Press.
Takahashi, T. 1989. ‘The influence of the listener on L2 speech’ in Gass et al. (eds) 1989a.
Takahashi, T. and L. Beebe. 1987. ‘The development of pragmatic competence by Japanese
learners of English’. JALT Journal 8: 131-55.
Tanaka, N. 1988. ‘Politeness: some problems for Japanese speakers of English’. JALT Journal 9:
81-102.
Tanaka, S. and S. Kawade. 1982. ‘Politeness strategies and second language acquisition’. Studies
in Second Language Acquisition 5: 18-33.
Tanaka, Y. 1991. ‘Comprehension and L2 acquisition: the role of interaction’. Unpublished
paper, Tokyo: Temple University Japan.
Tarallo, F. and J. Myhill. 1983. ‘Interference and natural language in second language acquisi¬
tion’. Language Learning 33: 55-76.
Tarone, E. 1977. ‘Conscious communication strategies in interlanguage: a progress report’ in H.
Brown et al. (eds.) 1977.
Tarone, E. 1978. ‘The phonology of interlanguage’ in Richards (ed.) 1978.
Tarone, E. 1979. ‘Interlanguage as chameleon’. Language Learning 29: 181-91.
Tarone, E. 1980a. ‘Some influences on the syllable structure of interlanguage phonology’. Inter¬
national Review of Applied Linguistics 4:143-63. Also in Ioup and Weinberger (eds.) 1987.
Tarone, E. 1980b. ‘Communication strategies, foreigner talk, and repair in interlanguage’. Lan¬
guage Learning 30: 417-31.
Tarone, E. 1982. ‘Systematicity and attention in interlanguage’. Language Learning 32: 69-82.
Tarone, E. 1983. ‘On the variability of interlanguage systems’. Applied Linguistics 4: 143-63.
Tarone, E. 1985. ‘Variability in interlanguage use: a study of style-shifting in morphology and
syntax’. Language Learning 35: 373-403.
Tarone, E. 1988. 'Variation in Interlanguage. London: Edward Arnold.
Tarone, E. 1990. ‘On variation in interlanguage: a response to Gregg’. Applied Linguistics 11:
392-400.
 Bibliography 785

Tarone, E., A. Cohen, and G. Dumas. 1976. ‘A closer look at some interlanguage terminology: a
framework for communication strategies’. Working Papers on Bilingualism 9: 76-90.
Tarone, E. and B. Parrish. 1988. ‘Task-related variation in interlanguage: the case of articles’.
Language Learning 38: 21^44.
Tarone, E., M. Swain, and A. Fathman. 1976. ‘Some limitations to the classroom applications of
current second language acquisition research’. TESOL Quarterly 10: 19-31.
Taylor, B. 1975. ‘The use of overgeneralization and transfer learning strategies by elementary
and intermediate students of ESL’. Language Learning 25: 73-107.
Taylor, D. 1980. ‘Ethnicity and language: a social psychological perspective’ in Giles et al. (eds.)
1980.
Taylor, D. 1988. ‘The meaning and use of the term “competence” in linguistics and applied lin¬
guistics’. Applied Linguistics 9:148-68.
Taylor, G. 1986. ‘Errors and explanations’. Applied Linguistics 7:144-66.
Taylor, L., A. Guiora, J. Catford, and H. Lane. 1969. ‘The role of personality variables in second
language behavior’. Comprehensive Psychiatry 10: 463-74.
Terrell, T. 1977. ‘A natural approach to second language acquisition and learning’. Modern
Language Journal 61: 325-36.
Terrell, T. 1991. ‘The role of grammar instruction in a communicative approach’. Modern Lan¬
guage Journal 75: 52-63.
Terrell, T., E. Gomez, and J. Mariscal. 1980. ‘Can acquisition take place in the classrrom’ in
Scarcella and Krashen (eds.) 1980.
Thakerar, J., H. Giles, and J. Cheshire. 1982. ‘Psychological and linguistic parameters of speech
accommodation theory’ in Fraser and Scherer (eds.) 1982.
Thomas, J. 1983. ‘Cross-cultural pragmatic failure’. Applied Linguistics 4: 91-112.
Thomas, L. and E. Elarri-Augstein. 1990. ‘The technology of learning conversations: a con¬
structive alternative to the mythology of learning styles’ in Duda and Riley (eds.) 1990.
Thomas, M. 1989. ‘The interpretation of English reflexive pronouns by non-native speakers’.
Studies in Second Language Acquisition 11: 281-303.
Thomason, S. andT. Kaufman. 1988. Language Contact, Creolization, and Genetic Linguistics.
Berkeley: University of California Press.
Thompson, E. 1991. ‘Foreign accents revisited: the English pronunciation of Russian immi¬
grants’. Language Learning 41: 177-204.
Thorndike, E. 1932. The Fundamentals of Learning. New York: Columbia Teachers College.
Tollefson, J. 1991. Planning Language, Planning Inequality. London: Longman.
Tomasello, M. and C. Herron. 1988. ‘Down the garden path: Inducing and correcting overgen¬
eralization errors in the foreign language classroom’. Applied Psycholinguistics 9: 237^46.
Tomasello, M. and C. Herron. 1989. ‘Feedback for language transfer errors: the garden path
technique’. Studies in Second Language Acquisition 11: 385-95.
Tomasello, M. and C. Herron. 1991. ‘Experiments in the real world: a reply to Beck and
Eubank’. Studies in Second Language Acquisition 13: 513-17.
Tomiyana, M. 1980. ‘Grammatical errors and communication breakdown’. TESOL Quarterly
14:71-9.
Tomlin, B. 1990. ‘Functionalism in second language acquisition’. Studies in Second Language
Acquisition 12: 155-77.
Tong-Fredericks, C. 1984. ‘Types of oral communication activities and the language they gener¬
ate: a comparison’. System 12: 133-34.
Tosi, A. 1984. Immigration and Bilingual Education. Oxford: Pergamon.
Towell, R. 1987. ‘Variability and progress in the language development of advanced learners of
a foreign language’ in Ellis (ed.) 1987a.
Tran-Chi-Chau. 1975. ‘Error analysis, contrastive analysis and students’ perceptions: a study of
difficulty in second language learning’. International Review of Applied Linguistics 13:
119-43.
786 Bibliography

Trueba, H., G. Guthrie, and K. Au (eds.). 1981. Culture and the Bilingual Classroom: Studies in
Classroom Ethnography. Rowley, Mass.: Newbury House.
Trevise, A. and R. Porquier. 1986. ‘Second language acquisition by adult immigrants: exempli¬
fied methodology’. Studies in Second Language Acquisition 8: 265-76.
Triandis, H. 1971. Attitude and Attitude Change. New York: Wiley and Sons.
Trubetzkoy, N. 1931. ‘Die phonologischen Systeme’. Travaux du Cercle Linguistique de Prague
4: 96-116.
Trudgill, P. (ed.). 1984. Applied Sociolinguistics. London: Academic Press.
Tucker, G., E. Hamayan, and F. Genesee. 1976. ‘Affective cognitive and social factors in second
language acquisition’. Canadian Modern Language Review 32: 214-16.
Tucker, G., W. Lambert, and A. Rigault. 1977. The French Speaker’s Skill with Grammatical
Gender: an Example of Rule-governed Behavior. The Hague: Mouton.
Turner, D. 1979. ‘The effect of instruction on second language learning and second language ac¬
quisition’ in Andersen (ed.) 1979a.
Turner, J. and H. Giles (eds.). 1981. Intergroup Behavior. Chicago, Illinois: University of Chi¬
cago Press.

Upshur, J. 1968. ‘Four experiments on the relation between foreign language teaching and learn¬
ing’. Language Learning 18: 111-24.
Uziel, S. 1993. ‘Resetting universal grammar parameters: evidence from second language ac¬
quisition of Subjacency and the Empty Category Principle’. Second Language Research 9:
49-83.

Valdman, A. (ed.). 1977. Pidgin and Creole. Indiana University Press.

Valdman, A. 1978. ‘On the relevance of the pidginization-creolization model for second lan¬
guage learning’. Studies in Second Language Acquisition 1: 55-77.
Valdman, A. 1992. ‘Authenticity, variation and communication in the foreign language class¬
room’ in Kramsch and McConnell-Ginet (eds.) 1992.
Van Buren, P. and M. Sharwood Smith. 1985. ‘The acquisition of preposition stranding by se¬
cond language learners and parametric variation’. Second Language Research 1: 18-26.
Van Els, T., T. Bongaerts, G. Extra, C. Van Os, and A. Janssen-Van Dieten. 1984. Applied Lin¬
guistics and the Learning and Teaching of Foreign Languages. London: Edward Arnold.
Van Lier, L. 1982. ‘Analysing interaction in second language classrooms’. Unpublished PhD the¬
sis, Lancaster: University of Lancaster.
Van Lier, L. 1988. The Classroom and the Language Learner. London: Longman.
Van Lier, L. 1991. ‘Inside the classroom: learning processes and teaching procedures’. Applied
Language Learning 2: 29-69.
Van Naerssen, M. 1980. ‘How similar are Spanish as a first and foreign language?’ in Scarcella
and Krashen (eds.) 1980.
Van Riemsdijk, H. 1978. A Case Study in Syntactic Markedness. Liss: Peter de Ridder Press.
Vander Brook, S., K. Schlue, and C. Campbell. 1980. ‘Discourse and second language acquisi¬
tion of Yes/No questions’ in Larsen-Freeman (ed.) 1980.
Vann, R., D. Meyer, and F. Lorenz. 1984. ‘Error gravity: a study of faculty opinion of ESL
errors’. TESOL Quarterly 18: 427-40.
VanPatten, B. 1990. ‘The acquisition of clitic pronouns in Spanish: Two case studies’ in Van-
Patten and Lee (eds.) 1990.
VanPatten, B. and T. Cadierno. 1993. ‘Explicit instruction and input processing’. Studies in Se¬
cond Language Acquisition 15, 225-43.
VanPatten, B., T. Dvorak, and J. Lee (eds.). 1987. Foreign Language Learning: A Research Per¬
spective. New York: Newbury House/Harper and Row.
VanPatten, B. and J. Lee. (eds.). 1990. Second Language Acquisition—Foreign Language Learn¬
ing. Clevedon, Avon: Multilingual Matters.
 Bibliography 787

Varadi, T. 1980. ‘Strategies of target language learner communication: message adjustment’. In¬
ternational Review of Applied Linguistics 18: 59-71.
Varonis,E. and S. Gass. 1985. ‘Non-native/non-native conversations: a model for negotiation of
meaning’. Applied Linguistics 6: 71-90.
Verhoeven, L. 1991. ‘Predicting minority children’s bilingual proficiency: child, family and insti¬
tutional factors’. Language Learning 41: 205-33.
Veronique, D. 1987. ‘Reference to past events and actions in narratives in a second language: in¬
sights from North African workers’ French’ in Pfaff (ed.) 1987b.
Verschuren, J. and M. Bertucelli-Papi (eds.). 1985. The Pragmatic Perspective. Amsterdam:
John Benjamins.
Vigil, F. and J. Oiler. 1976. ‘Rule fossilization: a tentative model’. Language Learning 26:
281-95.
Vincent, M. 1986. ‘Simple text and reading text Part 1: some general issues’ in Brumfit and Car¬
ter (eds.) 1986.
Vogel, T. and J. Bahns. 1989. ‘Introducing the English progressive in the classroom: insights
from second language acquisition research’. System 17: 183-94.
Von Elek, P. and M. Oskarsson. 1975. ‘Comparative method experiments in foreign language
teaching’. Department of Educational Research, Molndal (Gothenburg, Sweden): School of
Education.

Wagner-Gough, J. 1975. ‘Comparative studies in second language learning’. CAL-ERIC/CLL

Series on Language and Linguistics 26.
Wagner-Gough, J. and E. Hatch. 1975. ‘The importance of input data in second language ac¬
quisition studies’. Language Learning 25: 297-308.
Walter-Goldberg, B. 1982. ‘A stranger in Portugal: a study of Portuguese foreigner talk’. Term
Paper, Ed. 676, University of Pennsylvania. Cited in Larsen-Freeman and Long 1991.
Walters, J. 1979. ‘The perception of deference in English and Spanish’ in Yorio et al. (eds.) 1979.
Walters, J. 1980. ‘Grammar, meaning and sociocultural appropriateness in second language ac¬
quisition’. Canadian Journal of Psychology 34: 337-45.
Wanner, E. and L. Gleitman (eds.). 1982. Language Acquisition: the State of the Art. New York:
Cambridge University Press.
Wardhaugh, R. 1970. ‘The Contrastive Analysis Hypothesis’. TESOL Quarterly 4: 123-30.
Waterson, N. and C. Snow (eds.). 1978. The Development of Communication. New York:
Wiley and Sons.
Watson, J. 1924. Behaviorism. New York: Norton.
Weinert, R. 1987. ‘Processes in classroom second language development: the acquisition of
negation in German’ in Ellis (ed.) 1987a.
Weinreich, U. 1953. Languages in Contact. The Hague: Mouton.
Weinstein, C. and R. Mayer. 1986. ‘The teaching of learning strategies’ in Wittrock (ed.) 1986.
Wells, G. 1980. ‘Apprenticeship in meaning’ in Nelson (ed.) 1980.
Wells, G. 1985. Language Development in the Pre-school Years. Cambridge: Cambridge Uni¬
versity Press.
Wells, G. 1986a. The Meaning Makers: Children Learning Language and Using Language to
Learn. London: Hodder and Stoughton.
Wells, G. 1986b. ‘Variation in child language’ in Fletcher and Garman (eds.) 1986.
Wells, G. and M. Montgomery. 1981. ‘Adult-child interaction at home and at school’ in French
and McLure (eds.) 1981.
Wenden, A. 1983. ‘Literature review: the process of intervention’. Language Learning 33:
103-21.
Wenden, A. 1986a. ‘What do second language learners know about their language learning? A
second look at retrospective accounts’. Applied Linguistics 7: 186-201.
788 Bibliography

Wenden, A. 1986b. ‘Helping language learners think about learning’. English Language
Teaching Journal 40: 3-12.
Wenden, A. 1987a. ‘How to be a successful learner: insights and prescriptions from L2 learners’
in Wenden and Rubin (eds.) 1987.
Wenden, A. 1987b. incorporating learner training in the classroom’ in Wenden and Rubin (eds.)
1987.
Wenden, A. 1991. Learner Strategies for Learner Autonomy. New York: Prentice Hall.
Wenden, A. and J. Rubin (eds.). 1987. Learner Strategies in Language Learning. Englewood
Cliffs, N.J.: Prentice Hall.
Wenk, B. 1986. ‘Crosslinguistic influence in second language phonology: speech rhythms’ in E.
Kellerman and Sharwood Smith (eds.) 1986.
Wesche, M. 1981. ‘Language aptitude measures in streaming, matching students with methods,
and diagnosis of learning problems’ in Diller (ed.) 1981.
Wesche, M. and D. Ready. 1985. ‘Foreigner talk in the university classroom’ in Gass and
Madden (eds.) 1985.
Weslander, D. and G. Stephany. 1983. ‘Evaluation of English as a second language program for
Southeast Asian students’. TESOL Quarterly 1983: 473-80.
Westmoreland, R. 1983. ‘German acquisition by instructed adults’. Unpublished paper, Hawaii:
University of Hawaii.
Wexler, K. and P. Culicover. 1980. Formal Principles of Language Acquisition. Cambridge,
Mass.: MIT Press.
Wexler, K. and R. Mancini. 1987. ‘Parameters and learnability in binding theory’ in Roeper and
Williams (eds.) 1987.
Whitaker, H., D. Bub, and S. Leventer. 1981. ‘Neurolinguistic aspects of language acquisition
and bilingualism’. Annals of the New York Academy of Sciences 379: 59-74.
White, L. 1977. ‘Error-analysis and error-correction in adult learners of English as a second lan¬
guage’. Working Papers on Bilingualism 13: 42-58.
White, L. 1981. ‘The responsibility of grammatical theory to acquisitional data’ in Hornstein
and Lightfoot (eds.) 1981.
White, L. 1985. ‘The pro-drop parameter in adult second language acquisition’. Language
Learning 35: 47-62.
White, L. 1986. ‘Implications of parametric variation for adult second language acquisition: an
investigation of the “pro-drop” parameter’ in Cook (ed.) 1986.
White, L. 1987a. ‘Against comprehensible input: the input hypothesis and the development of
second language competence’. Applied Linguistics 8: 95-110.
White, L. 1987b. ‘Markedness and second language acquisition: the question of transfer’. Stud¬
ies in Second Language Acquisition 9: 261-86.
White, L. 1989a. Universal Grammar and Second Language Acquisition. Amsterdam: John
Benjamins.
White, L. 1989b. ‘The adjacency condition on case assignment: do learners observe the Subset
Principle?’ in Gass and Schachter (eds.) 1989.
White, L. 1990. ‘Second language acquisition and universal grammar’. Studies in Second Lan¬
guage Acquisition 12:121-33.
White, L. 1991a. ‘Adverb placement in second language acquisition: some effects of positive and
negative evidence in the classroom’. Second Language Research 7:133-61.
White, L. 1991b. ‘Argument structure in second language acquisition’. French Language Studies
1:189-207.
White, L. (forthcoming). ‘Universal grammar and second language acquisition: current trends
and new directions’ in Bhatia and Ritchie (eds.).
White, L. and P. Lightbown. 1984. ‘Asking and answering in ESL classes’. Canadian Modern
Language Review 40: 288-344.
 Bibliography 789

White, L., N. Spada, P. Lightbown, and L. Ranta. 1991. ‘Input enhancement and question
formation’. Applied Linguistics 12: 416-32.
White, L., L. Travis, and A. Maclachlan. (in press). ‘The acquisition of wh-question formation
by Malagasy learners of English: evidence for Universal Grammar’. Canadian Journal of
Linguistics.
White, M. 1992. ‘Teachers’ questions—form, function, and interaction: a study of two teach¬
ers’. Unpublished paper, Tokyo: Temple University Japan.
Widdowson, H. 1974. ‘An approach to the teaching of scientific discourse’. RELC Journal 5.
Widdowson, H. 1977. ‘Pidgin and babu’ in Corder and Roulet (eds.) 1977.
Widdowson, H. 1978. Teaching Language as Communication. Oxford: Oxford University
Press.
Widdowson, H. 1979a. ‘The significance of simplification’ in Widdowson 1979b.
Widdowson, H. 1979b. Explorations in Applied Linguistics. Oxford: Oxford University Press.
Previously published in Studies in Second Language Acquisition 1.
Widdowson, H. 1979c. ‘Rules and procedures in discourse analysis’ in Myers (ed.) 1979.
Widdowson, H. 1983. Learning Purpose and Language Use. Oxford: Oxford University Press.
Widdowson, H. 1989. ‘Knowledge of language and ability for use’. Applied Linguistics 10:
128-37.
Widdowson, H. 1990a. ‘Pedagogic research and teacher education’ in Widdowson 1990b.
Widdowson, H. 1990b. Aspects of Language Teaching. Oxford: Oxford University Press.
Wiese, R. 1984. ‘Language production in foreign and native languages: same or different?’ in
Dechert et al. (eds.) 1984.
Wilkinson, L. (ed.). 1982. Communicating in the Classroom. New York: Academic Press.
Williams, J. 1989. ‘Pronoun copies, pronominal anaphora and zero anaphora in second lan¬
guage production’ in Gass et al. (eds.) 1989a.
Williams, J. 1990. ‘Another look at yes/no questions: native speakers and non-native speakers’.
Applied Linguistics 11:159-82.
Williams, K. 1991. ‘Anxiety and formal/foreign language learning’. RELC Journal 22: 19-28.
Willing, K. 1987. Learning Styles and Adult Migrant Education. Adelaide: National Curriculum
Resource Centre.
Winitz, H. (ed.) 1981a. The Comprehension Approach to Foreign Language Instruction. Row-
ley, Mass.: Newbury House.
Winitz, H. (ed.). 1981b. Native Language and Foreign Acquisition. Annals of the New York
Academy of Sciences, 379.
Witkin, H. and J. Berry. 1975. ‘Psychological differentiation in cross-cultural perspective’. Jour¬
nal of Cross-Cultural Psychology 6: 4-87.
Witkin, H. and D. Goodenough. 1981. ‘Cognitive styles: essence and origins—field dependence
and independence’. Psychological Issues Monograph 51.
Witkin, H., O. Oltman, E. Raskin, and S. Karp. 1971. A Manual for the Embedded Figures Test.
Palo Alto, Calif.: Consulting Psychology Press.
Wittrock, M. (ed.). 1986. Handbook of Research on Teaching. (3rd Ed.). New York:
Macmillan.
Wode, H. 1976. ‘Developmental sequences in naturalistic L2 acquisition’. Working Papers on
Bilingualism 11: 1-13.
Wode, H. 1977. ‘The L2 acquisition of/r/’. Phonetica 34: 200-17.
Wode, H. 1978. ‘The LI vs. L2 acquisition of English negation’. Working Papers on Bilin¬
gualism 15: 37-57.
Wode, H. 1980. ‘Phonology in L2 acquisition’ in Felix (ed.) 1980a.
Wode, H. 1981. Learning a Second Language 1: an Integrated View of Language Acquisition.
Tubingen: Gunter Narr.
Wode, H. 1984. ‘Some theoretical implications of L2 acquisition research’ in Davies et al. (eds.)
1984.
790 Bibliography

Wode, H., A. Rohde, F. Gassen, B. Weiss, M.Jekat, and P. Jung. 1992. ‘LI, L2, L3: continuity vs.
discontinuity in lexical acquisition’ in Arnaud and Bejoint (eds.) 1992.
Wolfram, W. 1985. ‘Variability in tense marking: a case for the obvious’. Language Learning 35:
229-53.
Wolfram. W. 1989. ‘Systematic variability in second-language tense marking’ in Eisenstein (ed.)
1989.
Wolfram, W. 1991. ‘Interlanguage variation: a review article’. Applied Linguistics 12: 102-6.
Wolfson, N. 1976. ‘Speech events and natural speech: some implications for sociolinguistic
methodology’. Language in Society 5:182-209.
Wolfson, N. 1983. ‘Rules of speaking’ in Richards and Schmidt (eds.) 1983.
Wolfson, N. 1989a. Perspectives: Sociolinguistics and TESOL. Rowley, Mass.: Newbury
House.
Wolfson, N. 1989b. ‘The social dynamics of native and nonnative variation in complimenting
behavior’ in Eisenstein (ed.) 1989.
Wolfson, N. and E. Judd (eds.). 1983. Sociolinguistics and Second Language Acquisition. Row-
ley, Mass.: Newbury House.
Wolfson, N., T. Marmor, and S. Jones. 1989. ‘Problems in the comparison of speech acts across
cultures’ in Blum-Kulka et al. (eds.) 1989b.
Wong, H. 1934. ‘The best English: a claim for the superiority of received standard English’. Soci¬
ety for Pure English 39: 603-21.
Wong-Fillmore, L. 1976. ‘The second time around: cognitive and social strategies in second lan¬
guage acquisition’. Unpublished PhD thesis, Stanford University.
Wong-Fillmore, L. 1979. ‘Individual differences in second language acquisition’ in Fillmore et al.
(eds.) 1979.
Wong-Fillmore, L. 1982. ‘Instructional language as linguistic input: second language learning in
classrooms’ in Wilkinson (ed.) 1982.
Wong-Fillmore, L. 1985. ‘When does teacher talk work as input?’ in Gass and Madden (eds.)
1985.
Wong-Fillmore, L. 1992. ‘Learning a language from learners’ in Kramsch and McConnell-Ginet
(eds.) 1992.
Wu, Z. 1981. ‘Speech act—apology’. Los Angeles: ESL Section, Department of English, UCLA.

Yamazaki, A. 1991. ‘The effect of interaction on second language comprehension and acquisi¬
tion’. Unpublished paper, Tokyo: Temple University Japan.
Yorio, C., K. Perkin, and J. Schachter (eds.). 1979. On TESOL ’79. Washington D.C.: TESOL.
Yoshida, M. 1978. ‘The acquisition of English vocabulary by a Japanese-speaking child’ in
Hatch (ed.) 1978a.
Yoshioka, K. and T. Doi. 1988. ‘Testing the Pienemann-Johnston model with Japanese: a
speech-processing view of the acquisition of particles and word order’. Paper presented at
the 8th Second Language Research Forum, Hawaii, University of Hawaii.
Young, D. 1986. ‘The relationship between anxiety and foreign language oral proficiency
ratings’. Foreign Language Annals 19: 439-45.
Young, R. 1988a. ‘Variation and the interlanguage hypothesis’. Studies in Second Language Ac¬
quisition 10:281-302.
Young, R. 1988b. ‘Input and interaction’. Annual Review of Applied Linguistics 9: 122-34.
Young, R. 1989. ‘Ends and means: methods for the study of interlanguage variation’ in Gass et
al. (eds.) 1989b.
Young, R. 1991. Variation in Interlanguage Morphology. New York: Peter Lang.
Young, R. 1993. ‘Functional constraints on variation in interlanguage morphology’. Applied
Linguistics 14: 76-97.
Young, R. and C. Doughty. 1987. ‘Negotiation in context: a review of research’ in Lantolf and
Labarca (eds.) 1987.
 Bibliography 791

Yule, G. and D. McDonald. 1990. ‘Resolving referential conflicts in L2 interaction: the effect of
proficiency and interactive role’. Language Learning AO: 539-56.

Zobl, H. 1980a. ‘The formal and developmental selectivity of LI influence on L2 acquisition’.

Language Learning 30: 43-57.
Zobl, H. 1980b. ‘Developmental and transfer errors: their common bases and (possibly) differ¬
ential effects on subsequent learning’. TESOL Quarterly 14: 469-79.
Zobl, H. 1982. ‘A direction for contrastive analsysis: the comparative study of developmental
sequences’. TESOL Quarterly 16: 169-83.
Zobl, H. 1983a. ‘Contact-induced language change, learner-language, and the potentials of a
modified contrastive analysis’ in Bailey et al. (eds.) 1983.
Zobl, H. 1983b. ‘Markedness and the projection problem’. Language Learning 33: 293-313.
Zobl, H. 1984. ‘Cross-language generalizations and the contrastive dimension of the in¬
terlanguage hypothesis’ in Davies et al. (eds.) 1984.
Zobl, H. 1985. ‘Grammars in search of input and intake’ in Gass and Madden (eds.) 1985.
Zobl, H. 1986. ‘Word order typology, lexical government, and the prediction of multiple,
graded effects in L2 word order’. Language Learning 36:159-83.
Zobl, H. 1988. ‘Configurationality and the subset principle: the acquisition of “V” by Japanese
learners of English’ in Pankhurst et al. (eds.) 1988.
Zobl, H. 1989. ‘Canonical typological structures and ergativity in English L2 acquisition’ in
Gass and Schachter (eds.) 1989.
Zuengler, J. 1989. ‘Assessing an interactive paradigm: how accommodative should we be?’ in
Eisenstein (ed.) 1989.
Author index

Abbott, G. 58 Baca, L. 511

Abraham, R. 476, 478, 479, 501, 502, 503, Bachman, J. 516
504,534, 542, 549, 649,651 Bachman, L. 24
Acton, W. 519 Bacon,S. 203, 505, 649, 651
Adams, L. 598 Bahns, J. 137,139
Adams, M. 99 Bailey, C. 125,472, 475, 479, 480, 728
Adamson, H. 153 Bailey, K. 526, 573, 574, 583, 609
Adger, C. 166 Bailey, N. 91, 92,137, 145, 491, 655
Adjemian, C. 363, 416 Baker, C. 199,241
Africa, H. 219 Ball, P. 234
Akagawa, Y. 424, 649, 651 Banbrook, L. 590, 591
Alamari, M. 641 Bardovi-Harlig, K. 20, 70,163,166, 422,
Albrechtsen, D. 64, 66 427,631
Al-Haik, A. 495 Barnes, D. 587, 589, 717
Allen, J. 567,575 Barnes, S. 267
Allen, P.282,571,576,609 Bates, E. 31, 82, 350, 369, 374, 378, 697
Allwright, D. 570, 573, 583, 607, 609 Beck, M. 661
Allwright, R. 28, 565, 567, 573, 575, 577, Beebe, L. 22, 23, 25, 39, 71,120, 124,125,
585,586,592 127,129,133,146,147,148,157,158,
Alpert, R. 482 163,164,178, 179,180,182,188,189,
Altman, H. 472 190,211,212,234, 236,241,318, 662,
Amerlaan, A. 400 671,672,675
Anderson, E. 190 Bell, A. 122,124, 129
Anderson, J. 33, 350, 388, 393, 408, 410, Bell, L. 424, 633, 662
412,533,692,699 Bell, R. 67
Anderson, R. 32, 98, 117, 231, 333, 334, Bellack, A. 568
350,378,381,412,465,717 Bellugi, V. 77, 78
d’Anglejan, A. 24, 108, 215, 228, 242, 246, Benedict, H. 267
247,290,501,658 Bennett-Kastor, T. 268
Aphek, E. 534, 553, 557 Berent, G. 690
Arabski, J. 311, 312 Beretta, A. 603, 652, 653, 678, 681, 682,
Ard, J. 102, 428 683,684, 690, 691
Arthur, B. 254, 258, 288, 290, 674 Berko, J. 76
Asher, J. 489, 552, 570, 645 Berkowitz, D. 148
Aston, G. 262, 277 Berry, J. 502
Atkinson, M. 76, 77 Berwick, R. 436,473,515, 531, 597, 598
Au, K. 569 Bhardwaj, M. 331, 633
Au, S. 237 Bialystok, E. 18, 31, 33, 71, 295, 296, 297,
Austin, G. 500 355, 356, 359, 361, 363, 397, 399,400,
Austin, J. 160, 706 402,413,441,463, 501, 502,503, 541,
Avery, P. 258, 263,276, 596 551, 552, 556, 582, 608, 652, 654, 675,
685
 Author index 793

Bickerton, D. 14, 74, 125,126, 129,130, 105,106,115, 246, 302, 308, 310,311,
313,354 312,314,315,342, 343, 381, 614, 652,
Billmyer, K. 166 698,714,726
Bingaman, J. TL1 Busch, D. 518, 521
Birdsong, D. 441,487,488,493, 525,526, Butler, J. 93
673, 675, 676 Butterworth, B. 130, 394
Birnbaum, R. 91, 93 Butterworth, G. 40, 96, 99
Blau, E. 274 Bygate, M. 600
Bley-Vroman, R. 34, 75,106,107, 108,114, Byrne, J. 26, 210, 230, 234, 235, 241,242,
443, 444,449,452, 459,464,466, 647 709
Bloom, L. 47, 77
Bloomfield, L. 265 Cadierno, T. 646, 657
Blum-Kulka, S. 24,162,163,167,168,170, Camarata, S. 268
172,175,181,183,190 Campbell, C. 246, 290
Bodman, J. 166 Canale, M. 13,24,157, 165
Bofman, T. 20, 70,631 Cancino, E. 106, 230
Bogen, J. 675 Cancino, H. 22, 40,114,136,140, 246, 446,
Bohn, O. 88,331 491
Bongaerts, T. 400 Cantor, T. 360
Bonikowska, M. 164, 166 Carlisle, R. 145
Borker, R. 204 Carrell, P. 163,168,169,171
Bosworth, L. 518 Carroll, J. 36, 204, 473, 494, 495,496, 499,
Bourhis, R. 124 527,613,618,648,649
Bowen, J. 307 Carroll, S. 641
Bowerman, M. 381 Carter, E. 501,502, 505, 649, 651
Boyle, J. 202, 203,204, 241 Carton, A. 551
Braine, M. 81 Castagnaro, P. 524
Brazil, D. 568 Castanos, F. 598
Breen, M. 581, 608 Catford, J. 519
Briere, E. 613, 618 Cathcart, R. 170,173, 261, 569, 584, 594
Brindley, G. 103, 104, 386, 632, 636, 658 Cattell,R. 520,521,525
Brock, C. 590, 591 Cazden, C. 75, 78, 99, 230, 272, 332, 333
Broeder, P. 98 Celce-Murcia, M. 57, 58, 63, 67
Broen, P. 248 Cervantes, R. 275
Brooks, N. 299, 300 Chambers, J. 201
Brown, A. 541 Chamot, A. 37, 55, 68, 531, 533, 534, 535,
Brown, C. 245, 534 536, 538,540, 543, 544, 549, 550,554,
Brown, H. 230, 502, 519, 526, 556, 609 558,560,696, 723
Brown, P. 135,160, 161 Chan, J. 520
Brown, R. 74, 75, 76, 77,106,250,264, Chapelle, C. 500, 501, 502, 504, 506, 518,
275,288,597,716 519, 527
Brown, T. 554 Chastain, K. 64, 481
Brumfit, C. 266,598,691 Chatow, R. 205
Bruner, J. 500 Chaudron, C. 27,139, 254, 256, 274, 291,
Brunet, G. 515 349,449, 566, 567, 568, 581, 582,583,
Bub, D. 484 584, 585, 586, 592, 608, 609, 674
Buczowska, E. 630, 633, 635 Chenoweth, A. 262, 584
Budar, E. 614, 618 Cheshire, J. 128
Buhr, M. 295 Chesterfield, K. 534, 555
Burstall, C. 202, 203, 205, 228, 484, 489 Chesterfield, R. 534, 555
Burt, M. 14,19,29, 50, 54, 55, 56, 60, 61, Chi, M. 533
62, 63, 64, 70, 72, 75, 91, 92, 94, 95, 99, Chiang, D. 102
794 Author index

Chick, K. 184 Craik, F. 554

Chihara, T. 511, 613, 618 Croft, W. 415,420,462,465
Child, J. 499 Cronbach, L. 647, 648
Chomba, B. 219 Crookes, G. 36, 71,120,130,132,133,150,
Chomsky, C. 76 473, 509, 515, 516, 517, 527,590,591,
Chomsky, N. 1,12, 26, 38, 43, 81, 248, 300, 595,597,681,682
347, 353,415,429,436,438,444,460, Cross, T. 249,250,251,269
462,463,702, 710, 712, 718, 719,727 Crymes, R. 609
Chun, A. 262 Crystal, D. 77, 220
CILAR (Committee on Irish Language Culicover, P. 433
Attitudes Research) 200 Cummings, M. 164, 671, 672, 675
Clahsen, H. 21, 32,101, 103,105, 107,145, Cummins, J. 24,198, 223, 224, 225, 226,
290, 312, 313, 350, 382, 383, 384, 385, 227,240,241,242,282, 358,490,497,
387,412,454,466, 634, 638, 715 571,576,694,696, 709
Clark, E. 78, 79
Clark, H. 78, 79 Dagut, M. 68
Clark, J. 570 Dahl, M. 23,159,162,164,187,188,189
Clark, R. 87 Dahl, 0.417,421,582
Clarke, M. 291 Dakin, J. 299
Clement, R. 209, 210, 227, 237, 481, 512, Daniel, I. 632
513,515 Danielewicz, J. 131
Clifford, R. 353, 354, 373, 604 Dankoff, D. 313
Clyne, M. 252, 290 Danoff, M. 226
Coan, R.520 Darden, L. 676, 683
Cochrane, R. 486 Davies, A. 116, 221, 242, 603
Cohen, A. 13,166,175,176,177,190, 224, Davies, E. 65, 66, 72
397, 475, 530, 531, 534, 553, 557, 641, Day, R. 233,262,263, 501, 502, 504,585,
686 592,593,609,610,702
Cole, M. 214 de Beaugrande, R. 166
Coleman, L. 326 Debose, C. 252, 265
Collier, V. 241 de Bot, K. 130,131,663
Committee on Irish Language Attitudes Decamp, D. 76
Research 200 DeCarrico, J. 84
Comrie, B. 1, 415, 417, 418, 419,426,465 Dechert, H. 33, 85, 341, 345, 393,412, 508
Conrad, A. 217, 219 de la Torre, R. 570
Conrad, L. 274 Dent, C. 90
Cook, V. 34,102,106,108,243, 352,438, Derwing, T. 258, 275
439,452,456,463,465' de Saussure, F. 130
Cooper, R. 217, 219 de Villiers, J. 76,77, 106
Cooremann, A. 376 de Villiers, P. 76, 77,106
Cope, J. 480, 481 Dickerson, L. 120,125,133,143,144,145,
Coppieters, R. 485, 487, 525 146,157,365,655
Corder, S. P. 1,19, 28, 44, 47, 48, 50, 51, 52, Dickinson, L. 516
53,56, 67,68, 70, 72, 89,115,227, 327, Dietrich, R. 633
330, 336, 337, 351, 352, 399,403,409, Diller, K. 527, 569, 642, 663
581, 652, 670, 686, 688, 700, 701, 710, Dittmar, N. 31, 216, 285
714, 721,722 Dodd, C. 269
Coste, D. 663 Doi,T. 382,387,410
Coulmas, F. 85 Dornyei, Z. 513
Coulthard, M. 159,189, 568, 574 Doughty, C. 27,28,245,261,266,276, 596,
Courchene, R. 585 597, 598,599, 639, 657, 661, 663
Crago, M. 248 Douglas, D. 149
 Author index 795

Dowd,J. 148 Fterch, C. 33, 64, 66, 123,157, 171, 172,

Downes, N. 582 278,279, 295, 336, 344, 398, 399,402,
Dressier, W. 166 403, 410, 412, 475, 535, 600, 654, 674
Dudley-Evans, A. 255 Falodun, J. 595
Duff, P.597,598 Fanselow, J. 567, 575, 585
Dulay, H. 14,19, 29, 31, 50, 54, 55, 56, 59, Fantuzzi, C. 675, 680, 691
60, 61, 62, 70, 72, 75, 91, 92, 94, 95, 99, Farhady, H. 76
105,106,115,246, 302, 308, 310, 312, Fasold, R. 158
314, 315, 342,343, 381, 614, 652, 698, Fathman, A. 215, 216, 491, 613, 616, 618,
714,726 627, 628,635
Dumas, G. 322, 351, 397, 411, 686 Felix, S. 70, 96,101,106,117, 311, 360,
Dunkel, H. 514 436,438,443,455,463,466,621,622,
Duskova, L. 54, 55, 57, 62 624, 625,631,660
Ferguson, C. 245, 248,250, 252, 265,288,
Early, M. 582,589 290
Eckman, F. 322, 323, 335, 339, 345,417, Ferrier, L. 249
421,424,630,633,634,635,662,713 Fiksdal, S. 185
Edelsky, C. 208 Fillenbaum, S. 199
Edmondson, W. 172, 580, 581, 585 Fillmore, C. 89
Edwards, J. 200, 218 Finer, D. 451
Ehrlich, S. 258,263,276, 596 Finnemann, M. 203
Ehrman, M. 521, 540, 541, 542, 543, 545 Finocchiaro, M. 516
Eisenstein, M. 137,158, 166, 204, 211, 212, Fischer, J. 122, 157
213,615,619,650 Fishman, J. 217, 219
Ekstrand, L. 497 Fisiak, J. 309
Ellis, G. 556 Fitzpatrick, F. 223
Ellis, N. 360, 363, 642, 646 Flanders, N. 567
Ellis, R. 2,4,20,22,23, 31, 39, 81, 84, 86, Flanigan, B. 583, 710
87, 89, 90,106,116,120,133,135,136, Flege, J. 331,484
137,144,150,156,157,158,171,173, Fletcher, P. 116,290
184,189,190, 243, 262, 268,272,278, Flick, W. 61,302
283,285,288, 290, 302, 350, 361, 362, Flowerdew, J. 189
365, 374, 391, 395, 399, 412, 441,476, Flynn, S. 449, 453,465,718
479, 480, 501, 502, 505, 568, 577, 578, Foster, S. 461
582, 590, 592, 594, 603, 604, 606, 609, Fotos, S. 644, 646
610, 611, 616, 617, 619, 621, 624, 625, Fraser, B. 167,169, 175,176
629, 632, 643, 644, 645, 653, 654, 661, Fraser, C. 135
662, 663,672,681,685,686,688,698, Freed, B. 233,251,255
704 French, F. 48
Ely, C. 482, 511, 518, 592, 593 Freudenstein, R. 567
Enright, D. 581 Frohlich, M. 37, 501, 502, 503, 567, 575,
Epstein, N. 226 576, 608, 609
Erard,R. 519 Fromkin, V. 71
Erickson, F. 183,185, 569 Frota, S. 245, 361, 362
Ervin-Tnpp, S. 85,105,108,170,171, 315, Fry,J. 475
343 Fujimoto, D. 274
Eubank, L. 101, 452, 463, 466, 621, 625, Furrow, D. 267
661
Extra, G. 98 Gates, S. 254, 259, 569, 573, 582, 595, 609
Eysenck, H. 520, 525 Garcia, R. 489
Eysenck, S. 520 Gardner, R. 26, 36, 127,199, 200, 203, 208,
209,210,227,228,230,236,237,238,
796 Author index

242,475,476,481,482,483,494,495, Gremmo, M. 568,581

496,498, 509,510, 511,512, 513, 514, Griffiths, R. 37,254,255,274, 500, 501,
515, 516, 518,524, 526,527, 674, 680, 506, 517, 520, 522, 525, 527, 582, 681,
715,723 686
Garman, M. 116, 290 Gubala-Ryzak, B. 457
Garnica, O. 248 Gudykunst, W. 207,234
Gaskill, W. 262,290 Guiora, A. 517,518,519
Gass, S. 102,158,190, 203,216, 245,260, Gumperz, J. 183,184
263,264,266,272,278,289,290, 303, Gundel, J. 97
334, 339, 344, 345, 349, 376,377,424, Guthrie, G. 569
425,427,428,432,463,464, 465, 596, Guttfreund, M. 267
597, 599, 610, 630, 633, 656, 686, 691 Guy, G. 240
Gasser, M. 33,407,411
Gatbonton, E. 127, 135, 137,138,144, 365, Haastrup, K. 535, 594
366,705 Haber, R. 482
Gelman, R. 251 Hahn, A. 96,625, 631
Genesee, F. 25, 206, 226, 227, 241, 497, 503, Hakansson, G. 254, 255, 257, 582, 583, 588
650 Hakuta, K. 83, 84, 85, 86, 95,106, 272,446
George, H. 47, 61, 302 Hale, T. 614, 618
Gerbault, J. 446 Hall, B. 234
Ghambiar, S. 256 Halliday, M. 463, 578
Ghambiar, V. 256 Hamayan, E. 270, 271, 503, 582
Gibbons, J. 83, 84 Hammarberg, B. 68, 333
Gick, M. 642 Hammerley, H. 226, 241, 642
Gieve, S. 507 Hammond, R. 571, 602
Giles, H. 25,124,128,209,210,230, 234, Hanania, E. 83, 85, 88
235, 236, 241,242, 709, 724 Handscombe, J. 291, 609
Gillette, B. 546, 548 Hanlon, C. 250, 288
Gingras, R. 242 Hansen, J. 476, 503
Ginsberg, R. 516, 663 Hansen, L. 501, 502, 503, 504
Givon, T. 1, 31, 313, 369, 371, 372, 373, Hansen, N. 650
463 Hansen-Strain, L. 102, 424
Glahn, E. 581 Harder, P. 187
Gleason, J. 250 Harkness, S. 248
Gleitman, H. 249, 250, 267, 268 Harley, B. 282, 489, 491, 525, 571, 576,
Gleitman, L. 249, 250, 267, 268 622,626,637, 653, 662
Gliksman, L. 481, 511 Harri-Augstein, E. 473
Godfrey, D. 141 Harrington, M. 376-7, 378, 412
Goldin-Meadow, S. 289 Hartford, B. 163,166
Goldman-Eisler, F. 394 Hartnett, D. 613, 618, 649, 651
Goldstein, L. 211, 212, 241 Hatch, E. 21,23,27, 40, 76, 83, 85, 95, 96,
Gomez, E. 602 99,108,110,116,117,159,189,244,
Goodenough, D.501 251,252,253,254,255,256,259,264,
Goodluck, H. 435, 729 265,270, 284,285, 333,403,476, 508,
Goodnow, J. 500 600, 631, 648, 675, 680, 686, 687, 691,
Gradman, H. 83, 85, 88 693,704
Grauberg, W. 62, 302 Haugen, E. 306, 310, 343
Green, P. 359, 495, 500, 501, 502 Hauptman, P. 570
Greenberg, H. 415, 420 Hawkins, B. 277
Greenfield, P. 90 Hawkins, J. 415,418
Gregg, K. 23,119,155,278,356,429,437, Hawkins, R. 275, 425, 427, 448, 449, 465
675,676, 677,678,681,691 Heath, S. 206
 Author index 797

Hecht, K. 359 Ito, T. 376

Heidelberger Forschungsprojekt 217,218, Itoh, H. 83, 85
233,252
Heilenman, K. 375, 376 Jackson, H. 339
Hendrickson, J. 583 Jackson, K. 308
Henriksen, B. 64, 66 Jacob, P. 648
Henzl, V. 254,255,582 Jaeggli, O. 432, 446
Herbert, R. 164 Jain, M. 57
Hermann, G. 515 Jakobovits, L. 352
Hernandez, E. 50, 614 James, A. 116, 333
Herron, C. 639, 661 James, C. 40, 66, 308, 309, 344
Hewstone, M. 234 Janicki, K. 213
Heyde, A. 518 Jarvis, G. 570
Fliga, C. 254, 257, 259,261 Jefferson, G. 579, 584
Higgs, T. 353,354,373, 604 Johannson, S. 67
Hilles, S. 440, 446, 447,459, 464, 465 Johns, T. 255
Hinton, G. 404, 405 Johnson, D. 116, 691
Hirakawa, M. 450 Johnson, K. 647
Hirvonen, T. 290 Johnson, L. 490
Hodgson, R. 199 Johnson, P. 209, 274
Hoefnagel-Hohle, M. 27, 270, 271, 484, Johnston, M. 21, 32,103, 104,105, 114,
485,604 383, 385,386, 387,412, 636, 715
Holec, H. 499, 516, 568, 581 Johnston, S. 586, 589, 591, 608
Holliday, L. 283 Johnstone, R. 604, 608
Holmen, A. 581, 594 Jones, A. 419, 421, 424, 426, 630, 633, 660
Holmes, J. 164 Jones, C. 613,614, 616, 662
Holobrow, N. 25, 206, 241, 275 Jones, F. 597, 598
Holyoak, K. 642 Jones, S. 164
Horwitz, E. 472,478, 480, 481, 482,496, Jordens, P. 103,421,422
498,526 Judd, E. 40,190,216
Horwitz, M. 480, 481
Hosenfeld, C. 477, 544 Kachru, B. 213, 220, 221
House, J. 24,163, 167,168,170,172, 175, Kadia, K. 620, 621,625,653
190, 593 Kalema, K. 219
Howard, J. 608 Kamimoto, T. 305, 306, 343
Howell-Richardson, C. 482, 534 Kanagy, R. 595
Huang, J. 1, 21, 40, 83, 251, 285, 332 Kaneko, T. 578
Huang, X. 546, 548, 549 Karmiloff-Smith, A. 358
Huebner, T. 86,127,133,140,141, 313, Karp, S. 500
366,374 Kasper, G. 23,24, 33,159,160,161, 162,
Hughes, A. 64, 66 163,164,166,167,168,170,171,172,
Hulstijn, J. 68,124, 133,150,157,175, 357, 175,178, 183,187,188,189,190,278,
358,359,361,387 279,295, 317, 336, 340, 344, 398,399,
Hulstijn, W. 133,150,157, 359 402,403, 410, 412, 475, 535, 581, 585,
Hyams, N. 432, 439, 446, 460, 461,463 609,674
Hyltenstam, K. 102,103,139, 145, 303, Kassin, S. 360
320, 333, 345,421,425, 428,429,465, Kaufman, T. 310
647, 663 Kawade, S. 169, 171
Hymes, D. 13,157 Kay, P. 326
Kearsley, G. 587, 588
Ioup, G. 102,112,117, 443, 466,525 Keefe, J. 499
Issidorides, D. 275 Keenan, E. 284, 418,419, 426
798 Author index

Kelch, K. 274 Langacker, R. 407

Keller, J. 515 Lange, D. 101
Kellerman, E. 29, 62, 68, 300, 301, 303, 305, Lanoue, G.200
306, 308, 309, 311, 312, 316, 324, 325, Lapkin, S. 226, 566
326, 327, 328, 329, 330, 334, 335, 338, Larsen-Freeman, D. 3,13, 39, 40, 91, 93, 95,
341, 343, 344,345, 397, 398,400,401, 138,189,233, 252,270,271,278,289,
402,422,448,699, 721 290, 381,113, 387, 396,411,472,485,
Kelley, P. 233 609,611,627, 628, 631, 641, 662, 672,
Kemp, J. 205 681.690, 691
Khalil, A. 63, 65, 66, 701 Lascaratou, C. 64, 66
Kilborn, K. 376, 377 Laudan, L. 676
Kiparsky, C. 54 Laufer, B. 68
Klaus, G. 295 Lazaraton, A. 648
Kleifgen, J. 254, 257,290, 582 Leather, J. 116, 333
Klein, W. 12, 98,216,255, 285, 313, 331, Lee, J. 412
350,369,370,371,460 Lee, W. 48, 307
Kleinmann, H. 68, 481 Leech, G. 160
Kligell, R. 374 Legaretta, D. 582
Klima, E. 77, 78 Leino, A. 541
Koivukari, A. 587, 591 Lenneberg, E. 81,484
Konneker, B. 168, 169,171 Lennon, P. 69, 70, 71, 72, 215, 395, 546,
Kramarae, C. 202 548, 549
Kramer, A. 684 Le Page, R. 317
Kramsch, C. 291, 580, 663 Levelt, W. 130,131,132
Krashen, S. 14, 20, 27, 28, 31, 54, 55, 56, 72, Levenston, E. 304, 305, 421, 717
83, 84, 86, 87, 88, 91, 92, 93, 94, 95, 99, Leventer, S. 484
109, 117,133,190, 205, 214, 215, 247, Levinson, S. 159,160,161,188
250,273,278,280,289,290, 314, 349, Lewis, N. 283
355,356,359, 360, 363, 364,409,410, Lewis, S. 360, 567, 581
411, 482,484, 485, 491, 508, 549, 571, Liceras, J. 321
584, 602, 603, 613, 614, 615,616,618, Lightbown, P. 21, 97,113,114,246, 247,
638, 642, 644, 652, 655, 660, 661, 662, 270,271,290, 360,429,465, 572, 589,
663, 682, 692, 697, 698, 708, 712, 714, 603, 620, 621, 624, 627, 636, 637, 638,
723,726 640,641, 646, 655, 656, 660, 662, 672,
Krueger, N. 650 686,687,688,689,691,722
Kruidenier, B. 481, 512, 513 Lindberg, I. 588
Kucczaj, S. 407 Little, D. 478, 499
Kumpf, L. 145 Li, X. 280, 281
Kusudo, J. 570 Lockhart, R. 554
Lococo, V. 50, 62,139
Labov,W. 22, 121,122,123,124,125,129, Loewenthal, K. 489
130,147,157,202,206,580, 695, 716, Long, M. 3, 27,28, 35,39,67, 95,116,189,
724, 725, 728 233,245,252,253,254,256, 257,259,
Lado, R. 43, 306, 307, 310, 343 260,261,263,266,270,271, 273,274,
Lakshmanan, U. 272, 289, 446, 447, 463, 277,278,286,289,290,291,387, 396,
464 411,472,485,486,487,492,493, 526,
Lalonde, R. 209, 237 567,568, 573, 581,582,585,586, 587,
Lambert, W. 25, 27, 36,199, 203, 206, 208, 588,589, 590, 591,595,596, 597, 600,
209,218,228,241, 359,494,495,509, 607, 608, 609, 610, 611, 613, 616, 617,
510,514,518,715 639, 641, 656, 657, 662, 663, 667, 672,
Lamendella, J. 354 676, 677, 678, 681, 682, 683,684, 686,
Lane, H. 518,519 688.690, 691
 Author index 799

Lorenz, F. 66 McDonald, J. 375, 376

Lorscher, W. 579, 580 McGroarty, M. 37, 534, 542, 551, 552, 559,
Loschky, L. 276, 280, 281, 647 560
Lott, D. 59, 302 McHoul, A. 579, 580
Low, G. 132 McLaughlin, B. 2, 30, 32, 33, 88,106, 350,
Lowenberg, P. 220, 240 356, 359, 362, 378, 389, 393,408,410,
Lubin, J. 274 411,412,500, 543, 567, 648, 649, 650,
Lucas, E. 585 654, 674,681,683,691,722
Ludwig, J. 63, 203 McLean, M. 598
Lukmani, Y. 514 McLeod, B. 389,390,412
Luppescu, S. 262 McNamara, J. 105, 108, 215, 516
Lust, B. 449 McNeill, D. 76, 81,433
Lynch, A. 254 McTear, M. 568, 575, 577, 589
Lyons, J. 84,119 Meara, P. 112
Mehan, H. 568, 573, 574
MacIntyre, P. 476, 481, 482, 483, 498, 510, Meisel, J. 21, 32,103,105,145, 216, 252,
513,514 253,254,265,285, 290, 350, 382, 383,
MacIntyre, R. 496 384, 385,387,412,454, 617, 633, 634,
Maclachlan, A. 443 638,700,715, 722
MacPherson, J. 209 Meyer, D. 66
MacWhinney, B. 31, 82, 350, 369, 374, 378, Midorikawa, H. 591
405,407, 697 Miller, G. 248, 433
Madden, C. 91, 92,137,290, 491, 610, 655, MilonJ. 99,332
691 Mitchell, C. 164,171,172,176
Magnet, J. 223 Mitchell, R. 569, 604, 608, 691
Maier, M. 648 Mizon, S. 582
Major, J. 409 Mohatt, G. 569
Major, R. 147,316,330 Mohle, D. 33, 393,412
Makino, T. 627, 628 Montgomery, C. 615, 619
Malouf, R. 269 Montgomery, M. 261, 568
Maltz, D. 204 Moorcroft, R. 482
Mancini, R. 463 Morgenthaler, L. 283
Mangubhai, F. 534, 535, 552, 553, 559, 560 Morren, B. 535
Mannon, T. 274, 582 Morrison, D. 131
Manuel, S. 453 Moskowitz, G. 479, 567
Maple, R. 232 Muchnick, A. 510
Marchena, E. 68 Miihlhauser, P. 492
Mariscal, J. 602 Mukkatesh, L. 302, 353, 409
Marmor, T. 164 Muysken, P. 103, 312, 313, 454, 466
Martin, G. 613 Myhill, J. 322,425, 426
Martin, J. 307
Mason, C. 613, 618 Nagara, S. 313
Mateene, K. 219 Naiman, N. 37, 83, 503, 512, 517, 518, 519,
Matsumoto, K. 480 534, 535, 536, 546, 547, 549, 560, 592,
Mayer, R. 531 593, 648,705
Mazurkewich, I. 422, 447,449 Nakajima, K. 224, 490
McClelland, J. 32, 33, 348, 404, 405, 406, Naro, A. 265
407,411,717 Nation, P. 553
McConnell-Ginet, S. 291 Nation, R. 543, 648, 649
McCreary, J. 599, 644, 716 Nattinger, J. 84
McDonald, D. 600 Nayak, N. 650
McDonald, F. 604, 606 Nelson, D. 424, 633, 662
800 Author index

Nelson, K. 79, 267, 269, 508 Penfield, W. 484

Nemser, W. 309, 409, 710 Perdue, C. 369, 370, 371
Neufeld, G. 484, 486 Perkins, K. 190, 495, 497, 511, 627, 628,
Newmark, L. 17, 44, 314, 315, 652, 706 631
Newport, E. 249, 250, 251, 267, 268, 490 Perry, F. 554
Newton, J. 596,597 Peters, A. 79, 508, 693, 704
Nicholas, H. 137, 647 Petersen, C. 495
Nickel, G. 655 Pfaff, C. 141,369,372,373
Nishimura, M. 256 Phillips, S. 569
Nobuyoshi, J. 283,290, 604, 606, 661 Phinney, M. 447
Noyan, C. 633 Piaget, J. 347
Nunan, D. 566, 590, 595, 686, 689 Piatelli-Palmarini, M. 347
Nyikos, M. 203, 542, 545 Pica, T. 22,27,28, 75, 94, 95, 203,245,259,
Nystrom, N. 585, 586 261,266, 276, 279,283,291, 360,581,
582, 595, 596, 597, 598, 599, 620, 621,
Obler, L. 493, 495 624, 627, 628,631,653,662
Ochs, E. 131,248,288, 365, 718 Pickett, G. 549
Odlin, T. 29, 301, 302, 310, 312, 313, 317, Pienemann, M. 21, 32, 74, 88,101,103,104,
318,332, 341,343,344,711 105,114,115,145,290,350, 360,382,
O’Grady, W. 460 383, 384, 385, 386, 387,412, 629, 632,
Okamura-Bichard, F. 209 633, 634, 635, 636, 637, 638, 645,646,
Okuda, S. 85,88, 117, 272 653,654,655,662,663, 685, 715, 726
Oleksy, W. 190 Pierson, R. 237
Oiler, J. 200, 209, 241, 354, 495, 497, 511, Pimsleur, P. 36, 494, 495, 496, 498
584,613,618 Pinker, S. 407
Olsen, J. 584 Plann, S. 599
Olsen, L. 486 Pleh, C. 378,382
Olshtain, E. 161,166,168,170, 172,175, Plunkett, K. 404
176,177,178,182,188,190, 205, 306 Politzer, R. 37, 54, 534, 542, 543, 551, 552,
Oltman, O. 500 559,560, 567,570,581
O’Malley, J. 37, 533, 535, 536, 538, 540, Poole, D. 592
543, 544, 549, 554, 557, 558, 560, 696 Porquier, R. 285
O’Neill, W. 465 Porter, J. 106
Orem, R. 609 Porter, P. 266, 598, 599, 600, 610
Oskarsson, M. 642 Poulisse, N. 33, 340, 400, 401, 402
Oxford, R. 37, 480, 531, 532, 534, 536, 539, Prabhu, N. S. 595, 602, 652, 653
541,542,543,544,545,553,556,560 Prator, C. 213
Oyama, S. 484, 489 Preston, D. 124,125,127,130,135,138,
157,158,182,201,204, 205,213, 363,
Painchaud, G. 228, 658 366,368, 369,695,706
Palmer, A. 571 Prince, A. 407
Paribakht, T. 397 Purcell, E. 316
Parker, K. 27, 274, 275, 291,460
Parkinson, B. 482, 534, 604, 608 Quirk, R. 220, 221
Parrish, B. 141,147
Parry, T. 499, 527 Rabinowitz, M. 533
Pask, G. 500, 647 Ramage, K. 512
Patkowski, M. 484, 486, 489 Ramat, A. 372, 373
Pavesi, M. 70, 102,103,424, 629, 631, 635 Ramirez, A. 54, 567, 581, 586, 604
Pawley, A. 85, 88 Ramirez, J. 582, 589
PDP Research Group 33, 348, 404 Rampton, B. 124, 149,185,190
Peck, S. 244,251,259,284 Ranta, L. 21, 360
 Author index 801

Raskin, E. 500 Samuels, S. 486

Rathbone, M. 479,480, 619, 625 Santos, T. 65, 66, 67, 70
Raupach, M. 33, 341, 345, 393, 394,412 Sapon, S. 36, 495, 496
Ravem, R. 1,21,99,332 Sasaki, Y. 274, 376, 377
Ready, D. 254, 290,582 Sato, C. 67, 95,116,147,210,270, 271,
Reber, A. 359, 360, 642 350, 369, 372, 373,491, 568, 586, 587,
Regan, V. 153 588,589,591,610
Reibel, D. 314, 706 Satterly, D. 267
Reid, J. 506 Saunders, N. 143
Reiss, M. 546, 547, 548, 549, 560 Savignon, S. 515, 615
Renaud, C.228, 502, 658 Saville-Troike, M. 83, 84, 90, 116,708,717
Rescorla, L. 85, 88,117, 272 Sayegh, L. 275
Reynolds, P. 676 Scarcella, R. 84, 87, 88, 117,166, 169,
Richards, J. 44,55, 57,58, 59, 60, 71, 72, 172, 182,190, 254, 257, 259, 261, 485,
117,189, 217, 241, 242, 275, 689 486
Rieck, B. 98 Schachter, J. 19,23, 57, 58, 63, 67, 68, 72,
Rigault, A. 359 102,138,140,190,281, 304, 305, 308,
Riley, P. 187,190,568,581 316, 338, 345, 423, 443, 444, 452, 454,
Riney, T. 116,491 459,465,466,639, 641
Ringbom, H. 316, 328, 337, 341, 344 Schegloff,E. 579,584
Rintell, E. 164,166,169,171, 172,176 Scherer, A. 569
Ritchie, W. 442, 464 Schieffelin, B. 248, 288
Roberge, Y. 641 Schinke-Llano, L. 215, 216
Roberts, C. 184, 501, 504, 518, 519 Schlue, K. 246, 290
Roberts, L. 484 Schmidt, E. 245, 361, 362
Robinett, B. 72 Schmidt, M. 139, 421
Robertson, J. 91, 93 Schmidt, R. 31, 36, 40, 75, 95, 147,169,
Robson, G. 521, 525, 591 173,189,216,233,282,361,363,393,
Rodriguez, R. 83, 84, 225 395,396,403,404,407,409, 411,412,
Rogers, T. 44 413, 473, 509, 515, 516, 517, 527, 604
Romaine, S. 156 Schneider, W. 390
Rosansky, E. 22, 95,106, 136,140, 230 Schouten, M. 409
Rosenstein, D. 641 Schultz, J. 183,185
Rosing-Schow, D. 594 Schumann, F. 479, 516
Rossier, J. 516 Schumann, J. 22, 25, 99,102,136,140, 216,
Ross, S. 473, 515, 591 230, 231, 232, 242, 354, 372, 479, 516,
Rossman, T. 389, 390, 412 620,624,653, 681, 682, 683, 684, 691,
Rost, M. 278,591 692,721,723
Rubin, J. 475, 526,531, 532,533, 534, 535, Schwartz, B. 457
536, 546,547,559 Schwartz, J. 262
Rulon, K. 597,599, 644,716 Schwartz, R. 268
Rumelhart, D. 32, 348, 404, 405, 406, 407, Scollon, R. 161, 183,268,285
411,717 Scollon, S. 161,183
Rutherford, W. 34, 96, 312, 313, 316, 317, Scott, B. 500, 647
421,428,429,465, 631, 645, 656, 658 Scovel, T. 36, 354, 479, 482, 485, 488, 492,
Ryan, E. 209, 357 526
Scribner, S. 214
Sachs, J. 248 Searle, J. 160,167, 706
Sacks, H. 579,584 Seliger, H. 305, 359, 442, 474, 492, 501,
Saito, M. 464 502, 503, 526, 531, 532, 559, 592, 593,
Sajavaara, K. 340, 627, 628 608, 609, 610, 613, 618, 638, 642, 656,
Salica, C. 584, 585 657, 662
802 Author index

Selinker, L. 30, 44, 114,149, 309, 322, 344, Stenson, B. 60

349, 350, 351,353, 354, 355,409, 411, Stephany, G. 614, 619
416,417,694,703,710 Stern, H. 4, 37,197, 531, 535, 571, 572, 652
Shapira, R. 252, 253, 254 Stevick, E. 546, 559, 654
Shapiro, F. 582 Stockwell, R. 307
Sharwood Smith, M. 27, 71,112, 114, 278, Stone, M. 604, 606
279, 288, 296,301, 341, 345, 356, 361, Strain, J. 102, 424
434,449,654, 656, 661, 663, 698, 699 Strevens, P. 220
Shatz, M. 251, 268 Strickland, F. 519
Sheen, R. 37, 500, 502, 506, 525, 527 Stromquist, N. 586, 604
Sheorey, R. 65, 66 Strong, M. 515, 518, 520, 521, 527, 592,
Shiffrin, R. 390 593
Shimura, A. 305, 306 Sundland, D. 496
Shirai, Y. 675, 680, 691 Sundquist, L. 328
Shohamy, E. 205 Suter, R. 316
Silvius, W. 478 Svanes, B. 207, 209, 241
Simonot, M. 184 Svartvik, J. 50, 60, 71, 72
Sinclair, B. 556 Swain, M. 13, 24, 27, 157,198, 223, 224,
Sinclair, J. 568,574 226, 233,282,286, 322, 351,411,566,
Sinclair-de-Zwart, H. 81,115 571,576,594,602, 641,697
Singleton, D. 329, 345, 478, 489, 490, 492, Swan, M. 138
499,526 Syder, F. 85,88
Sjoholm, K. 316, 328
Skehan, P. 35, 36, 205, 472,474, 483, 495, Tabouret-Keller, A. 317
496,497,498,502,507,509, 517, 523, Tahta, S. 489
524, 525, 529, 535, 541, 558, 559, 591, Takahashi, T. 23,148,163,178,179,180,
648,651 182, 189,190
Skinner, B. 43, 299, 300 Takahiro 83, 84, 85
Skuttnab-Kangas, T. 222, 223, 224, 242 Tanaka, N. 170, 173
Slimani, A. 606 Tanaka, S. 169,171
Slobin, D. 32, 76, 77, 82, 379, 381, 382,408, Tanaka, Y. 276, 280, 281, 289, 606
717 Tarallo, F. 322, 425, 426
Smith, D. 231 Tarone, E. 22, 23, 31, 97, 112,116,119,
Smith, L. 220 125,127,133,138,140,141,142,144,
Smith, P. 570, 642, 697 145,146,147,148, 155,156,157,158,
Smythe, P. 228, 481, 511, 512, 515, 516 318, 350, 363,364, 365, 367,368, 397,
Snow, C. 27,188, 240, 248, 249, 251, 268, 398,401,403,409,412, 421,491, 530,
269,270,271,288,290,484,485, 604 641,655, 686, 687, 694, 725, 728
Snow, R. 647, 648 Taylor, B. 62, 330, 335, 609
Sokolik, M. 411 Taylor, D. 156,217
Sorace, A. 359, 442, 656,661 Taylor, G. 20, 57, 69, 72
Spada, N. 21, 97,360, 567,575, 576,603, Taylor, L. 519
609, 615, 619, 620, 624, 640, 641, 646, Terrell, T. 361,571,602, 652
688.691 Thakerar,J. 128
Spearritt, D. 648, 649 Thomas, J. 23, 52, 165,181,183, 719, 724
Spielberger, C. 479 Thomas, L. 473
Spolsky, B. 2,199, 203, 224,233, 515, 523, Thomas, M. 450
679.681.686.691 Thomason, S. 310
Sridhar, K. 220 Thompson, E. 487
Sridhar, S. 220 Thorndike, E. 299
Stansfield, C. 476, 501, 503, 527 Ting-Toomey, S. 207
Stauble, A. 233 Todesco, A. 37
 Author index 803

Tollefson, J. 214,238,242 Walters, J. 163, 168, 169, 171

Tomasello, M. 639, 661 Wardhaugh, R. 308
Tomiyana, M. 63, 64, 66 Waterson, N. 248
Tomlin, B. 31,369,373,410 Watson, J. 299
Tong-Fredericks, C. 597 Weigl, W. 622,625, 660
Tosi, A. 223 Weinbach, L. 166
Towell, R. 395 Weinberg, A. 531
Tran-Chi-Chau 302 Weinberger, S. 112, 117
Travis, L. 443 Weinert, R. 101, 621, 625
Trueba,H. 569 Weinreich, U. 306, 310, 343, 488
Trevise, R. 285 Weinstein, C. 531
Triandis, H. 199 Weintraub, S. 250
Trubetzkoy, N. 420 Weist, R. 630, 633, 635
Trudgill, P. 158,201 Wells, G. 76, 79, 80,116,250, 261,267,
Tucker, G. 228, 359, 503 268,269, 290,434,497
Tucker, R. 270, 271, 582 Wenden, A. 472, 477, 526, 534, 536, 537,
Turner, D. 360, 627, 628 541,556, 557,559,674
Wenk, B. 330
Uliss-Weltz, R. 163,178,179,180,190 Wertheimer, M. 569
Upshur, J. 613, 618 Wesche, M. 254, 290, 496, 527, 582, 649,
Usprich, C. 613, 614, 616, 662 650,663
Uziel, S. 444-5, 459, 466 Weslander, D. 614, 619
Westmoreland, R. 632
Valdman, A. 247, 655 Wexler, K. 433,463
Van Buren, P. 449 Whitaker, H. 484
Van Els, T. 54, 67 White, L. 13,21,27, 34, 39, 62, 82,119,
van Hout, R. 98 278,279,291, 321,345, 360,422,429,
Van Lier, L. 568, 575, 577, 578, 579, 580, 432, 434, 435, 436,438, 439, 440, 441,
585,586,591,607,609,690 442, 443,445, 447,448, 449, 450, 451,
Van Naerssen, M. 94, 546, 548, 549 452,454,455,456,457,458,463,464,
VanPatten, B. 412, 621, 625, 646, 657 465, 466, 589, 622, 627, 636, 637, 653,
van Riemsdijk, H. 422 657,658, 662, 663, 680
Vander Brook, S. 246, 290 White, M. 588, 589, 590,591
Vann, R. 65, 66, 476, 478, 479, 501, 504, Whitnam, R. 308
534, 542, 549 Widdowson, H. 13, 134, 156,157, 220, 288,
Varadi, T. 396, 397 366,410, 645, 652, 686, 688, 689, 727
Varonis, E. 203, 245, 260, 263, 264, 266, Wiese, R. 394
290,596,597,599,610 Williams, J. 141, 246
Verdi, G. 211 Williams, K. 482
Verhoeven, L. 224, 240 Willing, K. 499, 501,506,507, 527, 541,
Veronique, D. 141, 145 647, 648
Vigil, A. 511 Winitz, H. 645
Vigil, F. 354,584 Witkin, H. 500,501,502
Vincent, M. 257 Wode, H. 21,29, 88, 99,112,113,137,139,
Vogel, T. 137 330,332,333,421
Von Elek, P. 642 Wolfe, D. 510
Wolfram, W. 120, 125, 143,145,157
Wagner-Gough, J. 137, 244, 251, 253, 254, Wolfson, N. 40, 124,163, 164, 165,166,
270,284 167,184, 189,190
Walker, C. 533 Wong, H. 221
Wallace, R. 620, 624 Wong-Fillmore, L. 20, 87, 88, 266, 344, 535,
Walter-Goldberg, B. 252 536,544,554,569,605,610
804 Author index

Wood, M. 489 Young, D. 480,482, 526

Woodward, J. 675 Young, R. 27,133,135,137,143,151,152,
Wu, Z. 177 261,274,276,287,290
Yule, G. 600
Yamazaki, A. 276, 280, 281, 289, 606
Yates, A. 604, 606 Zelinski,S. 613,614, 616, 662
Yip, V. 444,459 Zobl, H. 29, 30, 35, 312, 313, 320, 321, 332,
Yorio, C. 190,258,263,276, 596, 609 345,421, 432, 450, 630, 633, 634, 720
Yoshida, M. 112 Zuengler, J. 120,129,133, 148,149,236,
Yoshioka, K. 382, 387, 410 241,318
Subject index
The Index covers the Introduction, Parts One to Seven, and the Glossary.
Entries are arranged in letter-by-letter alphabetical order, in which spaces between words are
ignored; ‘bilingualism’ is therefore listed after ‘bilingual classrooms’ and before ‘bilingual
settings’.
References to chapter notes are indicated by page and note number, e.g. ‘capacity 157n9’.
Glossary references are indicated by ‘g’.
ability 156,158nl0
fluid 500-1
language aptitude 496
procedural 348
rote learning 496
absolute universals 417
accent see pronunciation
acceptability, error gravity 66-7
accessibility hierarchy 418-19, 423-6,
660n2, 692g
accommodation theory 148,182
acculturation model 230-4, 692g
accumulated entities 96, 111, 631
accuracy
effects of formal instruction 617, 619-27
frequency hypothesis 269-72
achievement
distinction from aptitude 494
effect of age 488-90, 491-2
effect of anxiety 481-3
relation to learner participation 592-3,
594
relation to learning strategy 552-3
relation to motivation 511-13, 515
achievement strategies 33, 398, 403
acquired knowledge, v. learned knowledge
14,356, 364, 644, 654, 661nll
‘acquisition’
terminology 6
v. development 438
v. learning 14-15, 356, 362, 692g
acquisition device see language acquisition
device
acquisition orders see orders of development
acquisition phase 137, 366
acrolects 125, 126
ACT see Adaptive Control of Thought Model
actions, learner see learner strategies
action strategy 268-9
active learners v. passive learners 511-12
active task approach 535
activity type, classroom discourse 575-6
acts, classroom discourse 574. 575
Adaptive Control of Thought (ACT) Model
32,388-9, 410nll,692g
additional language 11, 219, 692g
additions, errors 56
additive bilingualism 208, 218, 223
address, classroom discourse 577-8
addressee factors 128-9, 160-1,173,318
see also interlocutor factors
Adjacency Condition on Case Assignment
464nl8
adult, definition 458
affective states 472, 479-83, 693g
affective v. cognitive feedback 584
age
effect on foreigner talk 257, 261
effect on fossilization 354
effect on strategy choice 35-6, 484-94
sociolinguistic aspects 201-2
agency 374-6
agreement, form-function mappings 374
AH see accessibility hierarchy
alcohol, effects 519
allocation, turn-taking 580
alternative questions 588
ambiguity, tolerance 518,519
analogy
overextension 59
806 Subject index
v. analysis 299-300
analysis
learner’s task 370
v. analogy 299
v. control of knowledge 357-8,400
analytical instruction 576
analytical v. gestalt strategy 79, 508, 693g
analytical v. synthetic learning style 507-8
analytic v. experiential teaching 571-2
analytic v. holistic strategies 401-2
analytic v. memory-orientated learners 498
animacy, form-function mappings 374,
376-7
anomie 199, 693g
anxiety 479-83, 524nl, 693g
apologies 174-8
a posteriori contrastive analysis 308
apperceived input 349
appropriateness 165
approximative systems 309, 409nl
a priori contrastive analysis 308
aptitude see language learning aptitude
Aptitude-Treatment-Interaction 647-52
archistrategies 401, 402
articulator 131
assessment strategies 683
associative stage 388-9, 533
asymmetrical patterns 322-3
ATI (Aptitude-T reatment-Interaction)
647-52
atomism, in modularity 680
attainment see achievement
attention-getters 249
attention to form 150, 546
attention to meaning 549
attention to speech 124,146, 149
Attitude Condition 680
Attitude Motivation Index 511
attitudes, learner 198-200, 207-10
audiolingual method 44, 569-70
auditory learning 506
authoritative interpretation 52
authority-orientated learning style 507
automatic knowledge v. controlled
knowledge 654
automatic processing v. controlled processing
390
automatic transfer 336
autonomous stage 389, 533
autonomy of grammar 438-9
avoidance 68, 304-5, 398, 693g
awareness, good language learners 549
babu 220
baby-talk see caretaker talk
backsliding 353, 694g
bad learning situations 231
balanced bilinguals 208
baseline data 245
basic interpersonal communication skills 24,
198,694g
effect of social class 206
interdependency principle, 224, 240n3
relation to aptitude 497
relation to extroversion/introversion 520,
521
basic language 221
basic variety 371
basilects 125,126
behaviour
learner see learner strategies
linguistic universals 420
behaviourist learning theories 299-300, 694g
of errors 60
of input 26-7, 243
of LI acquisition 81-2
beliefs, learners’ 477-9, 541
Beliefs about Language Learning Inventory
478
benign overgeneralizations 658
BICS see basic interpersonal communication
skills
bilingual classrooms, ethnographic tradition
569
bilingualism 208, 694g
interference 310
mother tongue maintenance 223
social and cultural milieu 236
bilingual settings 229
Bilingual Syntax Measure 50, 61, 91,106,
614
Birmingham school of linguists 568
‘black box’ 243,295, 565,608nl
borrowing 28, 336-7
linguistic 310
borrowing transfer 310, 313
see also language transfer
bottom-up processing 278-9,282
bottom-up strategies 678
boundary exchanges 574
bounding nodes 431
‘breken’ study 324-5, 328
BSM see Bilingual Syntax Measure
‘built-in syllabus’ 44

CA see contrastive analysis
CAH see contrastive analysis hypothesis
CALLA (Cognitive Academic Learning
Approach) 558
CALP see cognitive academic language
proficiency
Canonical Order Strategy 385
capability 146,156, 364, 694g
Capability Continuum 363-5
capacity 157n9
careful style 22,145-6, 318,364,654-5,
695g
careful writing 91
caretaker talk 247,248-51,267-9, 592,
695g
Carrot and Stick Hypothesis 509, 513-14
case, form-function mappings 374
casual style see vernacular style
categorical systematicity 123
causative motivation 36
CEFT (Child Embedded Figures Test) 525n4
change from above 205,367, 695g
channel capacity 438, 695g
child-directed language see caretaker talk
Child Embedded Figures Test 525n4
children
formulaic speech 86, 88
learning strategies 554—5
silent period 82—4
simplification 88-9
see also caretaker talk
Chomskyan tradition, variability 119,155
clarification 262, 283
clarification/verification, strategy 536
class, social 204-7
classroom discourse 573-81
classroom interaction 28, 502, 565-610
classroom organization, effect on acquisition
605
classroom-orientated research v. classroom
research 566
classroom pidgin 226
classroom process research 573, 695g
classroom research v. classroom-orientated
research 566
classroom setting 227-8,544
clinical elicitation 50, 670, 672, 690n5
closed questions 592,608n4, 695g
closed reasoning questions 587, 589
closed syllables 112
closed tasks 598
closing phase, classroom discourse 574
Subject index 807
clusters 432-3,463n7
COALA software package 636
code archistrategy see linguistic archistrategy
code-communication dilemma 652
code-mixing 28-9
code-shifting 343-4n8
code-switching 28-9,126, 696g
cognitive academic language proficiency 24,
198,696g
effect of social class 205
interdependency principle 224
native-speaker ability 525n3
relation to aptitude 497
relation to extroversion/introversion 520,
521
Cognitive Academic Learning Approach 558
cognitive accounts 30-3, 295-7, 347-413
cognitive feedback v. affective feedback 584
cognitive interactionist theories 243-4
cognitive modularity 439-40
cognitive processes, v. learner strategies 39n4
cognitive processing model 81-2
cognitive restructuring ability 501
cognitive stage, of skill-learning 533
cognitive strategies 37, 536, 538-9, 544,
696g
cognitive style 37, 696g
see also learning styles
coherence, norms 683
collaborative discourse 284-5, 287
collaborative meaning making 269
colonial languages 218-20, 240nl
COLT see Communicative Orientation to
Language Teaching
communication, cognitive accounts 392-403
Communicational Teaching Project 602-3,
652,653
communication breakoff 260
communication processes, v. learning
processes 30
communication strategies 16, 396^103, 696g
classification 33
interlanguage theory 351
LI transfer 314
v. production and learning strategies 530
communication transfer 336-7, 343^4n8
communicative classrooms 571, 602-4
communicative competence 13,165, 696g
communicative features 576
communicative instruction, v. traditional
instruction 571
communicative learning style 507
808 Subject index
communicative orientation 290nl7
Communicative Orientation in Language
Teaching 572, 575-7
communicative pressure, effect on
fossilization 354
communicative style 371
communities, focused and unfocused 317-18
comparative fallacy 75
comparative method studies 565, 569-73,
697g
compensation strategies 539
compensatory strategies 398
competence
definition 156,157-8nl0, 697g
transitional 409nl
v. performance 12-13, 71n7
v. processing 297nl
see also communicative competence;
linguistic competence; pragmatic
competence; proficiency
competence errors 58
Competition Model 31, 350, 373-8, 410n7,
697g
Competition Model (Felix) 455
complete access view, UG 453,454, 455, 456
complexification 352
complex implicational universals 418
complex variation space 367
compliments 165-6
composition 388
comprehended input 278, 349
comprehensible input 27, 273-80, 286,
289nl3,314, 339, 697g
comprehensible output hypothesis 27, 281-2,
590,697g
comprehension
caretaker talk 250-1
effect of input 690n2
effect of simplification 275
comprehension checks 259-60
comprehension errors 47
comprehension monitoring 550
comprehension questions 587
concatenative research 474-5, 476, 496, 524,
697g
conceptualizer 130, 132
conceptual strategies 33, 401, 402
concrete learning style 507
conditionals, language transfer 341, 344nll
confirmation 262, 283
confirmatory research 476
conflict validity 375
connectionism 406-8
connection strengths 39n4, 405-6
consciousness 355-63, 361-2, 399
consciousness raising 643-5, 661nl0, 698g
conscious strategies v. unconscious strategies
531-2
conservative grammars 436, 463n9
constitutive ethnography 568-9
content, effect on learner error 49
context 22,197, 698g
social 214-28
context-induced variation 142-9
contextual cues 280
contextualization cues 184-5
contextualization strategies 554
contrastive analysis 47-8, 306-9, 340, 698g
contrastive analysis hypothesis 29, 43,
306-12, 685, 698g
control
acquisition of 33
v. analysis 357-8, 400
v. knowledge 71n7, 296, 395
control-based communication strategies 400
controlled knowledge v. automatic
knowledge 654
controlled processing v. automatic processing
390
convergence 26, 128,182, 210
convergent thinkers 500
conversational analysis 244
conversational implicature 23
conversation maintenance devices 396
co-operation, strategy 398, 537, 557
core features, Universal Grammar 432
core goals, classroom discourse 577-8
core rules 319
correction
caretaker talk 250
definition 584
formal instruction 639—41
see also error treatment; feedback
correlational studies 551-3, 567
correspondence, norms 683
COS (Canonical Order Strategy) 385
covert errors v. overt errors 52,177
CRAPEL 568, 581
creative construction 19, 60,246, 677, 698g
creativity, morphological 401, 402
creoles 125-6, 698-9g
critical period hypothesis 35-6, 484, 486,
487,489

Cross-Cultural Speech Act Realization
Project 163, 175
cross-linguistic influence 301, 699g
cross-sectional error analyses 50, 55, 68
cross-sectional studies
developmental patterns 75-6, 77
L2 vocabulary acquisition 112
morpheme studies 91, 94,106
requests 169, 170,171
word order rules 105
crucial information 258
crucial similarity measure 332
CTP see Communicational Teaching Project
cues
availability and reliability 375, 378
contextual 280
form-function mapping 375-7
cultural differences 248-9, 559n3
cultural milieu 236-7
cultural stereotypes 184—5
culture shock 480
data, types 669-91
data collection
confirmatory research 476
illocutionary acts 162^1
input and interaction 244-6
learner preferences 506
learning principles 450
learning strategies 533-5
naturalistic research 475
UG-based studies 441
see also methodology
data-gatherers 476, 508
dative alternation 448-9
debilitating anxiety 482-3
decision-making, error treatment 585-6
declarative knowledge 32, 388, 410nl2,
699g
declarative stage 388
decoding device, PDP 406
decontextualization 119
decontextualized acts 188, 189nl0
deep structure 430-1
deep theories 677-8
Defence Language Aptitude Battery 495
deficit theory, bilingualism 223
degenerate input 248, 249, 433
deixis 23
delayed effect of formal instruction 621
deliberate actions 531
deliberate output 357
Subject index 809
denativization 379
depidginization 379
description, goal of SLA research 38
descriptive adequacy 429-30, 699g
descriptive tradition 116nl3
development, v. acquisition 438
developmental axis 384
developmental errors 58, 60, 61
developmental factors, constraint on transfer
315,329-34
developmental features see Multidimensional
Model
developmental patterns see patterns of
development
developmental sequences see sequences of
development
dialects 211-12, 321
idiosyncratic 53, 409nl
see also change from above
diary studies 245, 475, 480, 534
difference, contrastive analysis 306-9
‘different competence, same mechanisms’ 454
difficulty, contrastive analysis 300, 306-9
diffusion 366
directed attention 538
direct feedback 282
directives 79,173
direct refusals 179
direct requests 168
direct speech acts 160
direct strategies 539
discourse
classroom 573-81
collaborative, effect on acquisition 284-5
distinction from text 288nl
input 246,247-66, 288nl
planned v. unplanned see planned
discourse
transfer errors 316-17
discourse analysis 244, 566, 568
discourse completion questionnaires 164,
178,671,672, 700g
discourse completion tasks 163,175
discourse domain 149
discourse function 142
discourse management 257-60,289n7, 700g
discourse repair 257, 260-3, 289n7, 700g
discourse structure 263^1
display questions 587, 588-9, 590-1, 592,
608n4, 700g
distance see language distance
distortion, pedagogic discourse 581
810 Subject index
distributional bias principle 380
divergence 128,209-10, 525n2
divergent thinkers 500
domain
discourse 149
error analysis 70,71nl0
domains, universe of hypotheses 338
double-markings 172
downgrading
apologies 175
requests 167
downward convergence 128, 265
downward divergence 128, 265
drills, overdrilling 60
dual access view, UG 453,455,456
Dual Knowledge Paradigm 364
dual relevance of input 279
dynamic paradigm 125-7,129-30,133, 366
EA see error analysis
E-approach see externalized approach
Early School Personality Questionnaire 521
echoic functions, teachers’ questions 588
Ecological Validity Hypothesis 378
editing by feel 409-10n5
educational contexts 222-8, 229
educational settings 25, 214-16, 700g
effective students, good language learners
550
effort 511, 512
EFT (Embedded Figures Test) 525n4
elaboration
foreigner talk 254-5,256
interlanguage development 31
strategy 536, 540
elaborative modifications 275
elaborative simplification 265, 383-4, 700g
elicitation 50, 670, 672-3
Embedded Figures Test 525n4
embedding 370
‘embroidering’ 259, 264
emergent properties, PDP 406
empathy 518, 519
encoding device, PDP 406
enrichment theory, bilingualism 223
environmental factors, relation to age 492-3
environments 125
epenthesis 112,116nll, 253
epistemic functions, teachers’ questions 588
epistemic sources of error 58
error analysis 18-20, 47-72,139, 701g
error evaluation 48, 63-7, 701g
error gravity 66-7, 701g
errors 700-lg
classification 61
description 43-57
evaluation 63-7
explanation 57-63
identification 50-4
intralingual see intralingual errors
negative transfer 301-2
repair 262
error treatment 583-6, 701g
essential conditions, illocutionary acts 160
ethnic identity 207-10
ethnocentrism 199-200
ethnography 116nl3,164,206, 566,568-9,
701g
ethnolinguistic vitality 234
European Science Foundation Project in
Second Language Acquisition by Adult
Immigrants 98, 369, 370
exchanges, classroom discourse 574
execution phase 398
exemplar-based representation 410nl3
expansions, promoting development 269
experience see language learning experience
experiential instruction 576
experiential teaching v. analytic teaching
571-2
experimental elicitation 50, 670,672, 690n4
expert learners see good language learners
explanation, goal of SLA research 38
explanatory adequacy 429-30, 701-2g
explicit instruction 642-3, 661n9
explicit knowledge v. explicit knowledge 31,
349-50,355-63,441-2, 653, 702g
expressive learners v. referential learners 508
expressives, LI acquisition 79
extent, error analysis 70, 71nl0
external factors 16,17, 24-8, 354
externalized approach 415, 702g
external modification 167,172
external variation 127
extraction, markedness 320-1
extravergence 129
extrinsic motivation 36
extroversion/introversion 518, 520-2,
525n5,702g
Eysenck Personality Inventory 519
face-threatening acts 167,168,174,178
facilitating anxiety 482-3
facilitation

effect of instruction 654-7
effect of LI 302—4
factor analysis 61,506, 525n5
factual questions 587
false concepts 59
falsifiability 459, 683
falsification of behaviour 581
fast writing 91
feedback 583-4, 702g
for children 250,288n4,434-5
cognitive v. affective 584
corrective, in formal instruction 639—41
effect on fossilization 354
output hypothesis 282-3
feel, editing by 409-10n5
female/male differences 153,202-4, 545
field dependence/independence 37,500-6,
651,702g
fine-tuning 250,288n3
first language see LI
First Language Acquisition Theory 77
first occurrence 74
fixed expressions see formulas/formulaic
speech
fixed factors, individual learner differences 35
FLAT (First Language Acquisition Theory)
77
FLES (Foreign Language in the Elementary
School) 495
flexible use of learning strategies 550
fluency 394, 395
fluid ability 500-1
focused contexts 317, 343n6
focused meaning negotiation 283, 290nl6
focusers v. scanners 500
focus on forms v. focus on form 639-41,
661n5
follow-up moves 574-5
foreign accents see pronunciation
foreigner talk 26, 245, 247, 251-6, 288-9n6,
703g
foreignizing 343—4n8
foreign language acquisition 11-12, 703g
Foreign Language in the Elementary School
495
formal determinism principle 380
formal instruction 20-1, 383, 611-63, 703g
formal knowledge 188, 437
formal learning 108, 214, 215
formal practice, learning strategy 357, 551
formal properties of language 14, 38nl
formal reduction strategies 398
Subject index 811
form-function analysis 127, 703g
form-function mappings 374-5
form-function networks 31, 366
form-function relationships 140-2,437,
463nl0
form-function variation 135
formulaic knowledge 355-6
formulas/formulaic speech 20, 84-8,106,
272-3, 394-5, 703g
formulator 131,132
fossilization 309, 326, 353-5,409n2, 703g
fragile features 268, 289n9, 703g
framework goals, classroom discourse 578
free variation 126-7,135,136-8,157n4,
366, 704g
frequency, linguistic universals 420
frequency analysis 75, 704g
frequency hypothesis 269-73, 704g
friendship orientation 513
FT see foreigner talk
function, definition 142
functional grammarians 463nl0
functional hypothesis 143
functionalist models and theories 31,156,
369-78, 704g
functional planning, strategy training 557
functional practice 356-7, 549, 551
functional reduction strategies 398
Functional-Typological Theory 369
functions strategy 284
functor errors 66
Fundamental Difference Hypothesis 453,456
fused fields 500
fuzzy behaviour 459
‘garden path’ technique 640
gate-keeping encounters 187
GEFT (Group Embedded Figures Test)
500-2,506
gender differences 153,202-4, 545
general processing strategies 314-15
general theory 679-80
generative school, linguistic universals 34
genre, effect on learner error 49
gestalt learners, v. analytic learners 508
gestalt strategy v. analytical strategy 79, 704g
global errors 66, 67, 704g
goals
classroom discourse 577-8
of SLA research 38
good language learners 37, 517, 546-50,
648,650-2, 705g
812 Subject index
good learning behaviours 37
good learning situations 231
Governing Category Parameter 451
Government/Binding Model 429, 431, 449,
463n5
gradual diffusion model 127,137, 705g
grammar
acquisition, effect of age 489-90, 492
autonomy 43 8-9
focus on 112-13
mental 352
polylectal 126,127
grammar-translation method 569-70, 571
grammatical competence see linguistic
competence
grammatical errors 63, 66
grammatical function 142
grammatical input modifications 254-7
grammatical intuitions 363-4
grammaticality
effect of interaction 283
of interlanguage talk 266
grammaticality judgements 321-2, 705g
critical period hypothesis 487
metalingual knowledge 673—4
pronoun acquisition 97
Subset Principle 450
Universal Grammar 441, 443-4, 464nl5
grammaticalization 285, 371, 705g
grammatical rules, explicit and implicit
knowledge 359-60
grammatical sensitivity 496
Group Embedded Figures Test 500-2, 506
group work, classroom interaction 597, 599-
601
guidance, classroom discourse 577
‘guided’ acquisition v. ‘spontaneous’
acquisition 12
f
habits 19,29, 60,299-300
Harvard Study 77
heavy environments 125
Hebb rule 405-6
here-and-now principle 249, 259
hesitation phenomena 394
hierarchical constituent structure 264
hierarchical research 474, 476, 705g
hierarchies see accessibility hierarchy
hierarchy of difficulty 307
historical variables 238-9
Hogan Empathy Scale 519
holistic strategies v. analytic strategies 401-2
holists v. serialists 500, 647-8
homogeneous competence model 119, 363-4
horizontal variation 134, 705g
hypercorrection 205, 706g
hypocorrection 205, 706g
hypotheses, universe 338
hypothesis-testing 351-3, 706g
I-approach see internalized approach
ideational function, classroom discourse 578
identity hypothesis see L2=L1 hypothesis
Identity Scale 199
idiosyncratic dialects 53, 409nl
IDs see individual learner differences
IFS (Initialization/Finalization Strategy) 385
Ignorance Hypothesis 314, 706g
illocutionary acts 23, 160-83, 706g
illocutionary aspects of requests 167
illusive ideal native speaker 247
ILT see interlanguage talk
images 559nl
immersion education programmes 25,206,
225-7, 229, 241n4, 706g
immigrant interlanguage 217
implicational scaling 76,105,126,145,424,
706g
implicational universals 418, 420
implicature 23
implicit instruction 642-3, 661n9
implicit knowledge v. explicit knowledge 31,
349-50, 355-63, 441-2, 653, 706g
implicit learning 362,409n4
impossible errors 454
impression management 160,183-5, 706g
impulsive thinkers v. reflective thinkers 500
incidental samples, error analysis 49, 71n4
incomplete understanding 260
incomprehensible input 279
incorporation strategy 284, 706g
indicators 122, 205, 263, 706g
indirect feedback 282-3
indirect negative evidence 435
indirect refusals 179, 188n6
indirect requests 168
indirect speech acts 160
indirect strategies 539
individual learner differences 2,17,18, 35-7,
471-527, 707g
effect on strategy 540-3
induced errors 60, 707g
inductive language learning ability 496
ineffective students 550
inferencing strategy 551

informal learning 108, 214, 215
information
in discourse management 258-9
effect on comprehension 276
temporary see intake
information exchange tasks 599
information gap tasks 283
information processing
Cognitive Theory 32
PDP405
information-processing model 389-91, 533
informers, teachers as 227-8,241n5, 581
inhibition 299-300, 518, 519
initial hypothesis 352-3
Initialization/Finalization Strategy 385
initial state, Universal Grammar 433
initiating moves 574-5
inner circle 213
inner-directed learners 83, 708g
innovation 31,125
input 26-8, 243-91
comprehensible see comprehensible input
Parallel Distributed Processing 405
input discourse 246, 247-66, 288nl
input enhancement 661nl0
input frequency 269-73
input frequency-output accuracy studies 286
Input Hypothesis 27, 273, 614, 661nll,
708g
input text 246-7, 288nl
insertion errors 66
instructed language acquisition 12, 708g
instruction, formal see formal instruction
instructional phase, classroom discourse 574
instrumental motivation 513-14
instrumental orientation 509-10, 513-14
intake 349, 361, 708g
enhancement see consciousness raising
integration 349
integrative motivation 237, 509-13
integrativeness 512
integrative orientation 386, 509, 513
intelligence
learner-instruction matching 648
relation to aptitude 495, 497-8
intelligibility, error gravity 66-7
interaction 26-8, 28, 243-91, 502, 565-610
interactional acts 159, 708g
interactional approach, communication
strategies 396-8
interactional features of requests 167
Subject index 813
interactionally modified input 280,289-
90nl4, 600
interactional modifications 249-51, 257-60,
262-3, 279-80, 708g
interactional sociolinguistics 183
interaction analysis 566, 567-8, 571-2, 575,
608n2, 709g
interactionist learning theory 243-4, 709g
interactive input, relationship with
comprehension 274-7
interdependency principle 224, 709g
interest, intrinsic 515-17
interface hypothesis 654
interface position 356, 709g
interference 29, 47-8, 301, 310, 343n3, 709g
interference errors 58, 60, 61,302
Inter-Group Model/Theory 210, 230, 234-6,
238-9,709-1Og
interlanguage 16, 30, 44,114-15, 416-17
definitions 350-5, 710g
Interlanguage Structural Conformity
Hypothesis 417
interlanguage talk 247-8, 265-6, 599, 710g
inter-learner variability 79,113
inter-learner variation 134-5
interlingual errors 59
interlocutor factors 148,180,181,318
see also addressee factors
interlocutor scaffolding 285, 371, 372
Internal Cause Hypothesis 509
internal factors 16,18, 28-35, 354
see also ‘black box’
internalized approach 415, 710g
internal modification, requests 167
internal variation 127
international contexts 220-1, 229
interpersonal function, classroom discourse
578
interpretation 52
v. practice 645-6
interpretation tasks 657
interrogatives 77-8, 246-7
inter-speaker variation 121,126, 710g
inter-stage fluctuation 409-10n5
interviews, investigation of strategies 534
intonation, adjustment in caretaker talk 248
intra-learner variability 113-14
intra-learner variation 134-5
intralingual errors 58, 59, 60, 62-3, 302,
710g
intra-speaker variation 121, 126, 710g
Intrinsic Hypothesis 509
814 Subject index
intrinsic motivation 36, 515-17
introspection 475
see also think-aloud tasks
introspective techniques 245-6
introversion/extroversion 518, 520-2, 702g
involvement, good language learning 549
IQ, learner-instruction matching 648
IRF exchanges 575, 587
irritation, error gravity 66-7
islands of reliability 85, 86, 395
keyword-semantic strategy 554
keyword strategy 554
kinaesthetic learning 506
KIT A Study 372
knowers/informers, teachers as 227-8,
241n5, 581
knowing about language, strategy 537
knowledge
acquired v. learned 14, 356, 364, 644, 654,
661nl1
analysed v. unanalysed 357-8
controlled v. automatic 654
declarative see declarative knowledge
degrees 348
formal 188, 437
formulaic 355-6
implicit see implicit knowledge
metalingual see metalingual knowledge
Parallel Distributed Processing 405
potential 337
prior, inhibition from 299-300
procedural see procedural knowledge
relation to competence 156
v. control 71n7, 296, 395
knowledge-based communication strategies
400
LI * v. learned
developmental patterns 76-82
input and interaction 267-9
Universal Grammar 433-7
see also L2=L1 hypothesis; language
transfer
L2 acquisition, terminology and definitions 6,
11-15
L2=L1 hypothesis 105-9, 71 lg
Labovian paradigm 121-5,129-30,133
LAD see language acquisition device
Lancaster group 606
language acquisition device 24, 27, 33, 81,
433,71lg
language anxiety 480, 481
language appreciation lessons 661nl2
language-centred instruction 612-47
language classroom 227-8, 229
language distance, constraint on transfer 315,
327-9
language ego 517
language faculty 435-6
language learning aptitude 36, 494-9, 71 lg
ATI studies 650-1
effect on strategy 541
relation to field independence 502
socio-educational model 236
language learning experience
effect on learner error 49
effect on strategy 543
language learning strategies 530-1
language level, constraint on transfer 315,
316-17
language making capacity 32, 379
language module 680
language pedagogy 4-5,44, 686-8
procedures, relation to motivation 516-17
rules 656-7
language processing mechanisms 193-4
language teaching controversy 569, 642
language transfer 19, 28-30, 299-345
apologies 177-8
definitions 301, 71 lg
effect on pragmatic competence 187
interlanguage theory 351
refusals 180
v. morphological creativity 401, 402
language use see use
langue 130
lateralization of language function 484
leader-timed rhythm 330
learned knowledge see knowledge, acquired
learner attitudes 198-200, 207-10
learner errors 47-72
learner factors 18, 24-35, 148-9, 541-3
learner-instruction matching 502, 647-52,
71 lg
learner language 18-24, 43-72
learning styles 508
samples 49-50, 475-6
similarity with ungrammatical foreigner
talk 252-3
learner output see output
learners
attitudes, towards error treatment 584
 Subject index 815

beliefs 477-9, 541 linguistic theory 1-2

individual differences see individual learner
differences
participation 592-5
tasks 369-71
learner strategies 37, 712g
learner variation 382, 383-4
learning 712g
beliefs about 477-8
focus on 38
Monitor Theory 356
Parallel Distributed Processing 405
v. acquisition 362
learning experiences 473
learning-on-error model 375
learning opportunity, lack 354
learning outcomes 473, 573-4
learning principles 449-51
learning processes, v. communication
processes 30
learning strategies 314, 351, 473, 529-60,
712g
learning styles 473, 499-508, 541-2, 651
see also cognitive style
lects 125-6
left-branching languages 304, 343nl, 425
lessons, classroom discourse 574
level, of learner, effect on errors 49
levels, PDP 405
Levertov Machine 654
lexical errors 63, 66, 316
lexical phrases 84-5
lexico-semantics 324, 326
lexis, adjustments 248, 253
light environments 125
Linear Structural Analysis 237
linguistic archistrategy 401,402
linguistic borrowing 310
linguistic category taxonomy 54
linguistic competence 437-8, 712g
effect on pragmatic competence 187,
189n9
v. pragmatic competence 296
see also logical problem of acquisition
linguistic context 135,143-5
linguistic environments 125
linguistic mismatch argument 227
linguistics, symbiosis with SLA research 1-2,
686
linguistic strategies 33
linguistic task language, effects of
introversion 521
linguistic typology 417-29
linguistic universals 18, 33-5, 373, 415-66
literal translation 343^4n8
local errors 66, 71n9, 713g
local morphemes 383
local standard varieties 213, 220
lockstep settings v. small-group work 598-9
locutionary acts 160
logical problem of acquisition 34, 437,
452-3,460, 713g
logical problem of the data 673-4
longitudinal error analyses 50, 55, 68
longitudinal studies
developmental patterns 75, 76-7
free variation 136-7
L2 vocabulary acquisition 112-13
linguistic context 144
morpheme studies 95,106
natural v. educational settings 216
requests 169,171,173-4
sex differences 202
Machiavellian motivation 511
macro-monitoring 132
macro-planning 130
major information 258
male/female differences 153, 202-4, 545
management activities, classroom discourse
577
Marburg form, interaction analysis 567
marked choices 25, 212
markedness 713g
constraint on transfer 29-30, 315, 319-24
typological universals 420-8, 462n3
Universal Grammar 432-3, 447-9, 463n9
Markedness Differential Hypothesis 323,
634,713g
marked parameters 35, 39n6
markers 122, 713g
massive samples, error analysis 49
mastery 348
Matched Guise studies 199
matched instruction 647-52
matching
foreigner talk 265
learner’s task 370
maximum exposure argument 227
meaning
attention to 549
negotiation see negotiation of meaning
meaningful language 577
816 Subject index
meaning-orientated instruction 639, 66ln8
mean length of utterance 77,113, 248
mechanical language 577
mechanisms, transition theories 682
medium, effect on learner error 49
memory 108-9
memory-orientated learners 498
mental grammars 352
mentalist theories of language learning 714g
of errors 60
of input 26-7, 243
of LI acquisition 81-2
of L2 acquisition 33—4, 295-7
of strategies 531
mental mechanisms, relation to age 492-3 '
mental muscle model 493
mental processes, contribution to acquisition
16
mental representations, of ‘rules’ 32
mesolects 125,126
metacognitive strategies 37, 536-7, 538, 544,
714g
metalingual awareness 317
metalingual judgements 673-4
metalingual knowledge 54,150, 656-7,
661nl3, 714g
metalingual strategies 549
methodology
of classroom learning research 566-9
of developmental pattern studies 114-15
of illocutionary act studies 162-5
of individual difference research 474-7
of input and interaction investigation
244-6
of learning strategy investigation 533-5
of UG-based research 440-2, 458
microethnography 569
Micro-Momentary Expression test 519
micro-monitoring 132
micro-planning 130
mime 401
mimetic gesture 401
minimalist position 29, 309-15
mini theories of learning 477
minor information 258
miscommunication 260
misinformations, errors 56
misorderings, errors 56
mistakes 51, 58, 714g
misunderstanding 260
mixed pairings, classroom interaction 600-1
MLAT see Modern Language Aptitude Test
MLU see mean length of utterance
mnemonic association 553
modality reduction 183,189n8, 714g
model preferences 71n5
Model of Second Language Learning 31, 355,
356-9
Modern Language Aptitude Test 36, 495,
496,499, 650
modified input 247-66
modularity 439-40, 680-1
monitoring 131-2,133, 356, 364, 546, 714g
Monitor Model/Theory 31, 350, 355, 356,
364,682,685
monitor-over-users 508
monitor-under-users 508
monolingualism 208, 714g
morphemes 77, 78, 90-9,106,138-9, 383
morpheme studies 20-1, 90-6, 490-1, 631,
715g
morphological creativity v. transfer 401, 402
morphological errors 66
motherese see caretaker talk
mother tongue
effect on learner error 49
maintenance 223-4,229
motivation 36, 236-7, 508-17, 715g
distinction from aptitude 494-5
effect on strategy choice 542
relation to learning style 507
moves, classroom discourse 574
Multidimensional Model 32, 350, 382-8,
632,715g
multi-factor approach, to variation 151-3
multi-functionality principle 380
multilingualism 11, 715g
multiple cognitive mechanisms 464nl2
Myers-Briggs Type Indicator 542
national language see official language
contexts
native-like competence, critical period 35-6
native speaker, illusive ideal 247
native-speaker concept 212-13
native-speaker judges, error evaluation 63, 66
native-speaker proficiency 486-8, 489, 492,
525n3
native-speaker usage
illocutionary acts 162, 164-5,166,183,
188n3n4
input text 246-7,288n2
nativization 220
Nativization Model 378-82
 Subject index 817

Natural Approach 571 non-interactive input, relationship with

natural communicative language 520
natural contexts 216-21, 229
naturalistic discourse v. classroom discourse
580-1
naturalistic language acquisition 12,17,
38-9n, 715g
naturalistic research 475-6
natural languages 416
natural language use, data 670, 671-2,
690n3, 690n5
naturally occurring speech, illocutionary acts
163
naturalness factors 333
natural order of development 94,109-10,
360,652
natural settings 24-5, 214-16, 715g
negative evidence 279, 434-5, 456-7, 584
see also feedback
negatives
crucial similarity measure 332
development 99-101
form-function relationships 140
free variation 136-7
input text 247
LI acquisition 77-8
placement 421-2, 462n4
typological markedness 427
negative strategy 161
negative transfer 177-8, 300, 301-2
see also interference
negotiation, foreigner talk 265
negotiation of meaning 260—4, 289n8, 716g
effect on comprehension 276
focused v. unfocused 283, 290nl6
input/interactional modifications 279-80
through interlanguage talk 266
neurological mechanisms, relation to age
492-3
neuroticism 525n5
neutral value, markedness 462n3
new standard varieties 220
nexus questions 588
Nijmegen project 400-2
no-access view, UG 453, 454, 455-6
nodes, PDP 405
nonce errors 55
non-communication 260
non-communicative v. communicative
classrooms 571
non-engagement 260
comprehension 274-7
non-interface position 356, 716g
non-intervention 652
non-local morphemes 383
non-native-speaker judges, error evaluation
63, 66
non-participant ethnography 568
non-systematic variation see free variation
non-understanding 260, 263
normal interpretation 52
note-taking, strategy 536, 557
noticed input 349
noticing 361-2
noticing the gap 361
Noun Phrase Accessibility Hierarchy see
accessibility hierarchy
noun phrases, form-function relationships
141
Nuclear English 221
null subjects 432, 447, 464nl7
object extraction 444-5
obligatory occasion analysis 74-5, 91, 716g
observation
of classroom interaction 575, 608n3
of illocutionary acts 163,164
of learning strategies 533-4
Observer’s Paradox 122, 716g
Official English Movement 226
official language contexts 212-13, 217-20,
229
off-record acts 160
off-record feedback 262
omissions, errors 56, 61, 66
one-to-one principle 32, 380
on-record acts 160-1
on-record feedback 262
onset 387
opening phase, classroom discourse 574
open questions 587, 608n4, 717g
open reasoning questions 587, 589
open syllables 112,116nll
open tasks 597
operating principles 32, 82, 350, 378-82,
717g
opportunities for learning 354
optimal monitor users 508
oral narratives, form-function relationships
141
oral/situational approach 44
ordering, illocutionary acts 160
818 Subject index
orders of development 20-2, 73-117, 716g
orientation
distinction from motivation 509, 513
segregative and integrative 385-6
ostensive definition 401
other-correction 262
other-directed learners 83, 717g
other-repetition, discourse repair 261
outcomes 473, 573-4
output
comprehensible see comprehensible output
plus correction 281, 282
spontaneous v. deliberate 357
overall pattern 546, 559nl
overdrilling 60
overextension of analogy 59
overgeneralization 59, 351, 657-8, 717g
over-indulgence 305-6
overt errors v. covert errors 177
over-use 305-6, 717g
Parallel Distributed Processing 32-3, 39n5,
403-8, 409n4,411nl5, 717-18g
parameters 430-2, 440-1, 445-7, 464nl9,
718g
Parameter-setting Model 453-4,464nl9,
718g
paratactic speech 372
parents 228
see also caretaker talk
partial access view, UG 453, 454,455-6
partial understanding 260
participant ethnography 568
participants 135
participation, learners 592-5
participation index 580
passive learners v. active learners 511-12
past tense learning, PDP 406-7
pattern associator network 406
patterns 84, 718g
patterns of development 20, 73-117, 718g
PDP see Parallel Distributed Processing
Pedagogical Grammar Hypothesis 656
pedagogy see language pedagogy
Pennsylvania Project 570
perceptions 325, 361
perceptual theory of monitoring 132
performance 12-13, 71n7, 718g
Parallel Distributed Processing 404
variability 459
peripheral features, UG 432
peripheral rules 319
perlocutionary acts 160
persistence 511,512
personality 517-22, 542-3
phenomena, distinction from data 674-5
phonemic coding ability 496
phonological transfer 330
phonology see pronunciation
Phrase Structure Principle 34
pidginization 233, 379
pidgins see classroom pidgin; creoles
pied piping 422-3, 427, 718g
Pimsleur Language Aptitude Battery 36, 495,
496,650
pitch, caretaker talk 248
PLAB see Pimsleur Language Aptitude
Battery
planned discourse v. unplanned discourse
131,365, 718-19g
planned language use 82
planning, strategy 537
planning models 130-1
planning phase 398
planning time 131,132-3,150-1, 157n7
plans, relation to outcomes 573-4
plausible interpretation 52
plural -s, variation 151-3
politeness 160-1,162,171-3, 181, 189n7n8
polylectal grammar 126,127
positive evidence 434,457
positive strategy 161
positive transfer 29, 300, 302M, 719g
post-articulatory monitoring 132
postsystematic errors 56
potential knowledge 337
‘pouring back and forth’ 264
poverty of the stimulus 433-5, 719g
power, distribution among participants 261
practicality 683
practice
mechanism of change 32
v. consciousness-raising 643-5
v. interpretation 645-6
practising, strategy 539
pragmalinguistic errors 180,181
pragmalinguistic failure 23,165-6, 719g
pragmatic aspects of learner language 159-90
pragmatic competence 159, 296, 437, 719g
pragmatic errors 71 n6
pragmatic function 142
pragmatic mode 371-2
pragmatics 23^1, 719g
pragmatic transfer 180,181

Prague School 420
pre-articulatory monitoring 132
predictive framework 387
preference model 212-14, 679
premodified input 276, 289nl2
preparatory conditions, illocutionary acts
160
preposition stranding 321, 322, 422-3, 427,
719g
presystematic errors 56
primary processes 365-6
principle of perceptual saliency 143
principles
learning 499-51
Universal Grammar 430-1, 719-20g
prior knowledge, inhibition from 299-300
private speech 84, 720g
proactive inhibition 299, 720g
probabilistic systematicity 123
problematic communication 260-3
problematicity 399, 400
problematic overgeneralizations 657-8,
662nl4
problem-solving module 455
procedural ability v. learning 348
proceduralization 33, 388-9
procedural knowledge 388, 392, 410nl2,
720g
procedural skill 392-403
processes
learning, v. communication 30
primary and secondary 365-6
v. strategies 39n4,295
processing, v. competence 297nl
processing operations 382-8
processing strategies 384-5
pro-drop parameter 431-2, 441, 445-7,
463n6n7
production accuracy see accuracy
production errors 47, 71 n
production plans 399
production strategies 530, 720g
production theory of monitoring 132
production transfer see communication
transfer
proficiency 156,157n9,197-8, 720g
effects of formal instruction 613-17,
618-19
relation to learner participation 592-3,
594
relation to learning strategy 551-3
program-product comparisons 566-7
Subject index 819
Progressive Matrices 502
projection capacity 35, 432-3, 634, 720g
Projection Hypothesis/Model 432-3, 634-5,
636,660n3
pronominal copies 102,103, 116n8, 303-4,
418-19, 425
pronouns 96-9
form-function relationships 142
free variation 137
relative clauses 101-3, 303^4, 418-19,
425,632-3
pronunciation
critical period 35
developmental sequences 112,116n 12
effect of age 486-8, 489,492
effect of linguistic context 143
LI transfer 316
style shifting 122-3
property theories v. transition theories 681-2
proportion studies 60-1
prototypicality 315, 324-7, 721g
pseudo-communication 577, 589
pseudo-questions 587, 589
psycholinguistic context 135, 149-51, 721g
psycholinguistic markedness 325
psycholinguistic models 120,130-3, 398-
403
psycholinguistic sources of error 57, 58
psychological convergence 182
psychological distance 231, 232, 721g
psychological factors, effect on fossilization
354
psychological processing theories 148
psychometric research tradition 566-7
psychotypology 29, 315, 327-9, 72lg
public errors, correction 641
pull effect 655
pure education 652
pushdown 263
pushed output 27, 282, 283M, 590
quantification 114
quantity principles 410n6
quantity universal 373
question-formation, effects of instruction 623
questionnaires
on beliefs about language learning 478
on illocutionary acts 163, 164
on learner anxiety 481
on learning strategies 534
self-report 199
820 Subject index

see also discourse completion concatenative see concatenative research

questionnaires hierarchical see hierarchical research
questions methods see methodology
discourse management 259 theory-led 2, 607
teachers’ 586-92 research-then-theory approach 2, 77,474,
676-8
reaction to response 263 resilient features 289n9, 722g
real communication 577 responding moves 574-5
realization of language as a system 536 response matching 129
reasoning questions 587 responses 263
receptivity 299, 574 to stimuli 300, 694g
recreation continuum 352, 721g restrictive simplification 233,265, 383,
recurrent systemic errors 55 383-4, 411nl5, 722g
reductionism, in modularity 680 restructuring 115, 390, 410nl3, 722g
reductionist approach, theory evaluation restructuring continuum 330, 352-3, 377,
684-5 722g
reduction strategies 33, 398, 403 Resultative Hypothesis 509, 515
redundancy 275, 276 resultative motivation 36, 514-15
reference groups 25, 211, 721g retrieval strategies 398
referential learners v. expressive learners 508 revision, interlanguage development 31
referential questions 587, 588-9, 590-1, rhythm 248, 330
608n4,721g right-branching languages 304, 425
reflective thinkers v. impulsive thinkers 500 right roof constraint 442-3
refusals 178-83 risk-avoiding 403
registers 246, 247 risk-taking 403, 518
see also caretalker talk; foreigner talk; ritual of understanding or agreement 262
interlanguage talk role plays, illocutionary acts 163
regression, foreigner talk 265 roles, classroom 227-8, 241n5
regularities 364, 406 rote learning ability 496
regularization 119, 254, 255-6 rote questions 587
rejection stage 84 rough-tuning 250
relative clauses 101-3 routines 84, 722g
avoidance 304-5 routinization 390
errors 68 rule-based knowledge 355-6
facilitative effects of transfer 303^4 rule-based representation 410nl3
universal 418-20, 423-6 rule-created speech 87-8
relative pronouns see pronouns, relative rule formers 476, 508
clauses rule-orientated instruction 639, 661n8
relativization 418-19, 423-6 rules
relevance principle 381 grammatical, knowledge of 359-60
relexification principle 381 ignorance of restrictions 59
reorganization phase 366 incomplete application 59
repair mental representations 32
discourse see discourse repair of speaking 187
error treatment 584, 585 unlearning 34
repetition
caretaker talk 249 samples
discourse repair 261 classroom discourse 577
strategy 536 error analysis 49, 71n4
replacement phase 137,144 ‘scaffolding’285, 371, 372
requests 167-74 scanners v. focusers 500
research 676, 690n7 scene 135

schismic talk 575
SCS (Subordinate Clause Strategy) 385
secondary processes 365-6
second language see L2
second-verb ellipsis, variation 139
secret errors, correction 641
segregation 222-3, 229
segregative orientation 385-6
selective attention hypothesis 656-7, 722g
self-confidence 512-13
self-correction 262
self-directed learners, motivation 516
self-directing strategies 538
self-esteem 518
self-evaluation, strategy 538
self-management 538
self-repair, error treatment 586
self-repetition, discourse repair 261
self-report data 674
self-report questionnaires 199, 483, 519
self-selection, turn-taking 580
semantically contingent speech 269
semantically transparent grammar 326
Semantic Differential Technique 199
semantic errors 67
semantic formulas
apologies 175-8
refusals 179-80,188n6
semantic function 142
semantic processing (top-down processing)
278,282, 690n2
semantic prototypes 326
semantic simplification 88-90, 365
see also structural simplification
semantic strategy 554
semilingualism 208, 722g
sensitive period 492
sentence processing 378,410n9
sentences, derivation from utterances 119
sequence, turn-taking 580
sequences of development 20-2, 73-117,
382,383,660n3, 699, 723g
effects of instruction 624, 628-36
serialists v. holists 500, 647-8
settings 24-5, 135, 214-16
Universal Grammar 430
sex differences 153, 202M, 545
shallow theories 677-8
SHARP tradition 76, 115n3
signification of grammatical forms 645
sign language, resilient and fragile properties
289n9
Subject index 821
silent period 20, 82-4,106, 115n5, 723g
SILL (Strategy Inventory for Language
Learning) 537, 539
simple implicational universals 418
simplification
effect on comprehension 275
elaborative see elaborative simplification
foreigner talk 254-5, 256,265
restrictive see restrictive simplification
semantic 88-90, 365
structural see structural simplification
simplified varieties 221
simulations, of past tense learning 406-7,
411nl6
sincerity conditions, illocutionary acts 160
situational approach v. structural approach
570
situational context 135,145-9
situational factors, effect on strategy 543-5
situation-specific anxiety 479-80
‘skeletonizing’ strategy 259, 264
skill, procedural 392^103
skill-building hypothesis 280-1, 723g
skill development 358
skill-learning models/theory 32, 350,
388-92, 533
skill-learning strategies 530-1
‘slips of the tongue’ 47
small-group work 28, 597, 599-601
social/affective strategies 537, 539, 723g
social class 204-7
social contexts 214-28, 317-18, 343n7
social distance 231, 232, 723g
social factors 24-6,197-242
constraint on transfer 317-19
effect on fossilization 354
effect on strategy 543-5
inter-speaker variation 121-5
social goals, classroom discourse 578
social groups 130,148
social interactionist theories 244
socialization, through caretaker talk 250
social marking 149
social milieu 236-7
social models 230-9
social psychological models 127-9, 133
social questions 587
social strategies 37, 555
Socio-educational Model of L2 Learning 26,
230,236-9, 723g
sociolinguistic aspects of requests 168
sociolinguistic competence 165
822 Subject index
sociolinguistic models 366-8
of variability 121-30,133
sociolinguistics, interactional 183
sociolinguistic sources of error 57-8
sociolinguistic variables 25, 724g
sociopragmatic errors 181
sociopragmatic failure 23,165, 724g
specific samples, error analysis 49, 71 n4
Speech Accommodation Theory 127-9,129,
724g
speech acts 159,160-2, 188nl,724g
speech comprehension system 131
speech maintenance 128
speech modification see teacher talk
speech monitoring models 133
speech planning 33,133, 393-6, 724g
speech production model 132
speech rate, effect on comprehension 274
‘spontaneous’ acquisition v. ‘guided’
acquisition 12
spontaneous generalization, PDP 406
spontaneous output 357
spreading 364-5
stability 367
standardization 119
state anxiety 479-80
statement-picture complex 682
status
discourse repair 261
learner, effect on pragmatic competence
187
steady state, Universal Grammar 433
stereotypes, cultural 184-5
stimulus, poverty of 433-5
stimulus/response 300
storage 349
strategic plans 399
strategic transfer 336
strategies 37, 529-60
for apologies 175-6, 188n5
communication see communication
strategies
involved in transfer 314-15
learning, interlanguage theory 351
linguistic 33
processing 384-5
for requests 168
v. processes 39n4, 295
v. tactics 289n7, 531-2
v. techniques 531
Strategy Inventory for Language Learning
537,539
strategy training 402-3
stress, form-function mapping 376
structural approach v. situational approach
570
structural predisposition for transfer 321
structural simplification 88-90, 115n6,
724-5g
structural variables 238-9
structure
linguistic universals 420
transfer 59
style shifting 123-5,129,135,145-9, 725g
stylistic continuum 22, 725g
stylistic factors, intra-speaker variation
121-5
subjacency 430-1, 442-5, 464nl3
subject extraction 444-5
subliminal learning 361
submersion 25, 224-5, 229, 725g
Subordinate Clause Strategy 385
Subset Principle 436, 449-51, 463n9, 657
subsidiary transfer 336
substitution function 284
substratum transfer 310, 313
subtractive bilingualism 208, 218
superordinate style see careful style
superset grammars 436
surface strategy taxonomy 55-6
surface structure 430-1
syllables, open and closed 112,116nll
symmetry 326
synchronic variation 125
syntactical features, effect of linguistic
context 145
syntactical structures, LI compared with L2
106-7
syntactic mode 371-2
syntactic processing (bottom-up processing)
278-9,282
syntactic structures 99-104, 111
syntactization 371-3
see also grammaticalization
syntax, caretaker talk 248
synthesis, learner’s task 370
synthetic learning style v. analytic learning
style 508
systematic data 672-3
systematic errors 56
systematicity 123
systematic variation 135,138-51,144
tactics v. strategies 289n7, 531-2

tactile learning 506
‘talking down’ 252,254,265
target language
illocutionary acts 162,164-5,166,183,
188n3n4
input text 246-7, 288n2
target-like use analysis 75, 94,115n2, 725g
target variety 211-14
task-based syllabuses 595
task design, effect on interaction 28
task frameworks 595-6
task-induced variation 135,138—40, 725g
tasks 595-7, 608n6
v. task requirements 157n8
task variables, classroom interaction 595-7,
598
teachability hypothesis 634, 655-6, 726g
teacher-led lessons, interaction 28
teachers, training 556-8, 591
teacher’s paradox 580
teachers’ questions 586-92
teacher-student relationships 227-8, 241n5
teacher talk 26, 581-3, 726g
teaching
through caretaker talk 250
through foreigner talk 264-5
see also language pedagogy
teaching exchanges 574-5
technical predispositions 536
techniques v. strategies 531
temporal variables 255, 394
tendencies, universal 417-18
text, distinction from discourse 288nl
textual function, classroom discourse 578
thematic structure, elaborative modifications
275
theories 676-86
theory-led research 2
Theory of Second Language Learning 685
theory then research 607
therapy, error evaluation 71n8
think-aloud tasks 475, 534-5, 726g
third language acquisition 11
time, relation to linguistic context 144
time pressure, effect on variation 150
TL see target language
top-down processing 278,282, 690n2
top-down strategies 678
topicalization 452
topic-comment structure 313,316-17
topics
caretaker talk 249
Subject index 823
control in discourse management 258-9
conversation-building with children 269
discourse domain 149
turn-taking 579
Total Physical Response (TPR) 552, 571, 645
traditional instruction v. communicative
instruction 570, 571
trailer-timed rhythm 330
training
learning strategies 556-8
teachers’ questions 591
training sets 405, 410-1 lnl4
training transfer, interlanguage theory 351
trait anxiety 479-80
transactions, classroom discourse 574
transfer see language transfer
transfer errors 59-63, 66, 726g
transfer hypotheses 97-8
transfer load 343—4n8
transfer of structure 59
transfer to somewhere principle 381
transfer of training, interlanguage theory 351
transitional competence 409nl
transitional competence errors 59
transitional constructions 726g
‘transitional forms’ 47
transitional stages 112
transitional structures 99, 111
transition theories v. property theories 681-2
translatability 327,328
translation
communication transfer 28
literal 343—4n8
travel orientation 513
treatment, of errors see error treatment
tribal model 678
trigger 243,263,461
turn-taking 579-80
tutor talk 583
see also interlanguage talk
type-token ratio, caretaker talk 248
typological universal 34, 320, 415-16,
417-29, 726-7g
UG see Universal Grammar
unattended speech data 146
unconscious learning 361
unconscious v. conscious strategies 531-2
unfocused contexts 317-18
unfocused meaning negotiation 283
ungrammatical input 252-4,272
unique errors 60, 61
824 Subject index
uniqueness fallacy 402
units, PDP 33, 404-5
Universal Grammar 34, 415-16, 429-66,
727g
markedness 319
relation to formal instruction 660nl
universal hierarchy 418
universalist position, word order 312
universals see linguistic universals;
typological universals
universal tendencies 417-18
see also negatives, placement
universe of hypotheses 338
unlearning 34, 299
unmarked choices 25, 212
unmarked parameters 35, 39n6
unplanned discourse see planned discourse
unplanned language use 82
unrestricted universals 417-18
‘unwanted forms’ 47
uptake chart 606
upward convergence 128
upward divergence 128
usage 13-14, 727g
use 13-14, 477, 727g
data 670
types, classroom discourse 577-8
U-shaped behaviour 303, 343nl, 727g
Utilitarian English 221
utterance processing 378, 410n9
validity of data 674-6
VARBRUL 124,125,152,153
variability 22-3, 119-58, 690n3
theories 350, 363-9
variability hypothesis 654-5, 727g
Variable Competence Model 365-6, 677,
685,727-8g
variable forms 121, 728g
variable rules 123-5,157n2,n3
variants 121, 201, 728g
see also frequency analysis
variation see inter-speaker variation; intra¬
speaker variation; learner variation
variational axis 384
variational features 105, 116n9
see also Multidimensional Model
varieties
creoles 125-6
official language 219-20
target 211-14
verbal intelligence, relation to aptitude 495
verb forms, variation 137
verb inflection, pro-drop parameter 446
verbosity 172, 182-3
vernacular style 22, 31,122,145-6, 318,
364,367, 654-5, 728g
vertical constructions 268, 285, 728g
vertical variation 75, 134, 728g
visual learning 506
vocabulary learning strategies 553-4, 556-7
voice contrast 322-3
VORD 499
waffle phenomenon 172
Wave Theory 125, 728g
weights 31, 39n5, 404
well-established findings 678-9
Wickelfeatures 406
wild grammars 435, 454, 463n8, 729g
word associations 112
word order
form-function mappings 374, 376-7
rules 103-5,112, 384-5, 631-2
transfer 312-14
written language, decontextualized 188
wrong choice errors 66
x-questions 588
York Language Aptitude Test 495
zero option 652-4
Zweitspracherwerb Italienischer, Spanischer
und Portugiesischer Arbeiter (ZISA) 20,
103-4,382,383,617, 700g
^OXFORD
APPLIED
LINGUISTICS

The Study of Second Language Acquisition

This book serves as a comprehensive, up-to-date, and
accessible introduction to second language acquisition
research and as a reference book for this important field
of study. The first section outlines a general framework
for the study of second language acquisition.
Subsequent sections provide a general description of
learner language, account for the role of the linguistic
environment, examine the learner’s internal
mechanisms, explore individual differences in language
learning, and review the expanding research on
classroom second language acquisition.

Rod Ellis is Professor of Teaching English as a Second

Language at College of Education, Temple University,
Philadelphia.

Applied Linguistics Adviser: H.G. Widdowson